Evaluating a Mobile Application Based Intervention for Insomnia in College Students: A Preliminary Study

Abstract

Objective:

This study examined the feasibility, acceptability, and efficacy of a self-guided, mobile application for Cognitive Behavioral Therapy for Insomnia (CBT-I Coach) in a sample of college students.

Participants:

Data was collected from 55 students, who mostly identified as women (82%) and white (84%) and reported at least moderate insomnia symptoms based on the Insomnia Severity Index.

Methods:

Participants were randomized to either an intervention condition (i.e., 4-weeks of CBT-I Coach) or a wait-list condition and completed self-report measures biweekly across the 8-week study period.

Results:

Nearly 70% of participants found the app moderately to extremely effective. The intervention group experienced a larger reduction in insomnia symptoms from baseline to post-treatment compared to the control group (g=0.88).

Conclusions:

These findings provide preliminary evidence that utilizing a self-guided mobile intervention for insomnia among college students is feasible and components of the app were perceived to be moderately to highly effective.

The overall prevalence of insomnia among college students is high, with a mean prevalence rate of 18.5% (range = 9 – 38%)¹. This has important implications on the physical and mental well-being of college students, as well as their academic performance. For example, students who report elevated insomnia symptoms are more likely to report greater levels of depression, anxiety, and stress ^2,3. Previous research has also shown that students who experience poor sleep quality obtain lower academic scores ^4,5. Taken together, these findings support that insomnia is a relevant and important issue among college students and that greater efforts to identify interventions that are compatible with this population are needed.

College Life and Sleep

The transition to college can be stressful. Students often go through numerous changes to their environments (e.g. moving to attend college), interpersonal relationships (e.g. decrease in communication with family/friends), and schedules during the transition to college ⁶. These stressors often predict and perpetuate insomnia symptoms among college students ⁷. Previous research has shown that anywhere between 8% to 13% of college students report difficulties falling and staying asleep for at least three nights a week ^8–10. This is even more important considering that inadequate sleep has the potential to impact attention, decision making, memory, daytime functioning and academic performance ¹¹. In addition, students who suffer from insomnia symptoms are at greater risk for developing depression and anxiety. For example, according to one study, approximately 78% of college students who suffer from insomnia also endorsed depressive symptoms ². As these students are graduating and entering the workforce, the continuation of insomnia symptoms may be contributing to higher costs to the healthcare system and employers. It is estimated, for instance, that insomnia is associated with a reduction of up to 54 days of workplace productivity a year. This garners losses of up to $207.5 billion in gross-domestic product ¹². Similarly, the indirect and direct healthcare costs of insomnia are estimated at $100 billion dollars per year ¹³.

Using Cognitive Behavioral Therapy (CBT-I) in College Students

The clinical implications are that early intervention efforts among college students may potentially reduce the risk of developing more chronic or severe forms of insomnia. Cognitive Behavioral Therapy for Insomnia (CBT-I) is the first line and “gold-standard” treatment for insomnia. It is a multi-component behavioral intervention that typically requires at least 6–12 treatment sessions lasting approximately 60 minutes each ¹⁴. In most cases, this intervention is highly effective, with 70–80% of patients experiencing improvements in their sleep following CBT-I ¹⁵. Treatment outcomes can include less time falling asleep, fewer awakenings during the night, and longer time spent asleep ¹⁶. A meta-analysis conducted in 2018 highlighted that CBT-I can also reduce symptoms of depression ¹⁷. For example, one study found that four sessions of CBT-I conducted by a licensed therapist significantly reduced depression and insomnia symptoms compared to a self-help group. The symptom reduction found in the CBT-I group was maintained at the three-month follow-up with rates of depression remission being 10 times greater than the other participant group ¹⁸.

However, CBT-I has several practical limitations that can limit the utility of this treatment, especially among individuals with less severe symptom profiles or who are otherwise relatively healthy (e.g. college students). That is, CBT-I is typically only available from clinicians who have specialized training in behavioral sleep medicine, time intensive and cost prohibitive, and commonly believed to be reserved for individuals with severe insomnia symptoms. College students who may be seeking medical attention for poor sleep have easier access to university health centers than they do behavioral sleep specialists ¹⁹. While this particular issue is not limited to college students, it does make access to CBT-I difficult in this population. Another limitation is the time commitment needed from a patient to complete a standard course of CBT-I (weekly or biweekly sessions for at least 2–3 months). Students may be unable or unwilling to participate in the course of treatment recommended by their healthcare provider due to busy academic schedules or other factors. There is also the consideration of the cost for a patient who has been prescribed CBT-I. Many students may not be able to afford insurance coverage or copays required to access CBT-I (it is often the case that behavioral sleep interventions are not covered by insurance) or may experience lapses in coverage when enrolled ²⁰. Lastly, CBT-I is often seen as treatment solely for chronic insomnia despite research showing that people who suffer from other sleep continuity disturbances or those with comorbid conditions may benefit from this intervention ^21–24. These limitations may have an impact on overall treatment uptake, especially among college students.

Alternative Intervention Options

As the internet has expanded, so has the opportunity for treatment dissemination and accessibility. In the case of internet-based or digital intervention, strategies often include tasks such as watching videos, recording symptomology, reading about or learning new skills, and tracking progress. Digital treatment options range from therapist guided interventions to entirely automated formats and from web-based software to mobile applications. Digital CBT-I (dCBT-I) is a promising and cost-effective alternative to traditional CBT-I. Originally, dCBT-I was used by clinicians as supplemental resources for patients outside of the clinical setting. However, the improvement of app content has opened the possibility of using app-based treatment without the need of a clinician (i.e., self-guided). Several dCBT-I applications have been created and distributed to healthcare and public consumer markets. It is important to note that not all available apps are created equally in terms of accessibility and scientific support. For example, platforms such as Somryst and Sleepio require a prescription from a health care provider or insurance company. For patients who tried to access Somryst without insurance, their out-of-pocket costs were approximately nine hundred dollars, and required a diagnosis of chronic insomnia from a licensed medical provider (Note: as of April 2023, Pear Therapeutics filed for bankruptcy and Somryst is currently no longer available for purchase) ²⁵. In contrast, other applications such as Sleeprate are missing cognitive components of CBT-I which is considered a cornerstone of insomnia intervention ²⁶. However, there are other app-based CBT-I interventions developed by the United States Veterans Affairs (VA) Department (i.e. CBT-I Coach and Insomnia Coach) that have the potential to increase accessibility to treatment by being available without prescriptions and having no cost to use ^27,28.

Previous research has shown significant decreases in insomnia symptom severity in study samples after using self-guided digital interventions for insomnia ³. Grierson et al. ²⁹ found that participants had significantly improved insomnia symptoms and psychological distress scores after engaging in a course of self-guided dCBT-I. Using self-guided dCBT-I has also been shown to be efficacious in at least one preliminary study aiming to reduced insomnia symptoms among nurses ³⁰. The sample consisted of seventeen female nurses who worked day shifts at a VA hospital and had a score of fifteen or higher on the Insomnia Severity Index (ISI). During the baseline visit, the nurses were required to download the CBT-I Coach app and complete training on how to use it. The nurses were encouraged to use the sleep diary, tools, and sleep prescription features of the app for a total of 6 weeks. The primary outcome measure (ISI scores) was evaluated at baseline (mean = 18.1), week 3 (mean = 12.8), and week 6 (mean = 11.5). Analyses of the nurses ISI scores resulted in a significant decrease in symptom severity from baseline to week 3 (mean ISI difference = 5.2) and from baseline to week 6 (mean ISI difference = 6.6). While this study did have a number of limitations (e.g. small sample size and no control group) it did provide initial support for the efficacy of self-guided dCBT-I and the CBT-I Coach app. Similar results were found in a study conducted by Kuhn et al. ²⁸. Fifty U.S. veterans (58% men) with moderate insomnia symptoms were randomly assigned to a self-guided CBT-I app (i.e. Insomnia Coach) or a wait-list control group for a six-week period. Participants completed sleep diaries and self-report measures at baseline, posttreatment, and a twelve-week follow-up. Approximately 75% of the study sample engaged with the app for the full six-week intervention period. Treatment effects were seen during the follow up in areas of insomnia severity (d = −1.1), sleep onset latency (d = −0.6), and depression symptoms (d = −0.8). These outcomes support the idea that self-guided digital CBT-I is a promising intervention option for improving insomnia symptom severity and other sleep-related outcomes.

The Current Study

An effective mobile version of CBT-I may increase the accessibility of behavioral interventions for sleep to populations that are often missed by healthcare providers. The goal of this study was twofold. The primary aim of this study was to quantify the effectiveness of self-guided dCBT-I in treating insomnia symptoms among college students. We predicted that (1) the intervention group, compared to a waitlist control group, would experience a significantly greater reduction in insomnia symptoms after participating in four weeks of a self-guided behavioral intervention for insomnia (i.e., CBT-I Coach), and (2) the waitlist control group would experience similar reductions in insomnia symptoms during the second, four-week period (when they received the active intervention). A second aim, considering recent literature suggesting that CBT-I has a positive effect on depressive symptomatology, was to evaluate the effect of this intervention on post-treatment depressive symptoms for both groups. Finally, we assessed the overall feasibility and acceptability of using a CBT-I app to treat insomnia symptoms in college students. The potential benefits of this project include: (1) self-guided dCBT-I can act as a more accessible treatment alternative to traditional CBT-I and/or hypnotics for students who are engaging in poor sleep hygiene practices. (2) dCBT-I could possibly reduce the cost of treatment for students who do not have insurance or cannot afford to pay high copays. (3) dCBT-I is accessible on a smartphone or tablet. Having therapy available on these devices would reduce the time students would have to budget to get to and from their therapy appointments, as well as the time engaged in the actual session with their provider.

Method

Participants

Participants included students from a large Midsouthern university who were interested in receiving insomnia treatment via a mobile app. They were recruited through their classes and through an online subject database. The online subject database was the result of a comprehensive Qualtrics survey that screened for potential sleep disorders and other risk factors. Class recruitment was conducted for four consecutive academic semesters (Fall 2021 through Spring 2023), in which general psychology students completed a prescreener at the beginning of each term. The prescreener included the Insomnia Severity Index (ISI). Of those students that completed the prescreener during these semesters, 480 students with a score of 15 or greater were invited via email to participate in this study. A score of 15 or greater on the ISI is commonly considered the clinical cut-off for clinical insomnia³¹. A total of 80 students were recruited to participate in the current study. All 80 students completed the baseline visit. However, 20 students were lost to follow up from baseline to week four, four students had ISI scores that were less than 8 during the baseline visit (i.e., their insomnia had remitted since the time they completed the prescreener survey), and one student reported having narcolepsy (see Figure 1). Our primary analyses therefore included data from 55 students. The differences between the excluded and included participants were analyzed. There were no differences between the groups in age (F (1, 78=0.834, p > 0.20, M included = 19.3 years, M excluded = 19.9 years), gender (χ²= 0.02, p > 0.20, 82% women included, 72% women excluded), race (χ²= −0.11, p = 0.11, 84% white included, 64% white excluded), employment (χ²= −0.04, p = 0.19, 71% unemployed included, 52% unemployed excluded), baseline ISI scores (F (1, 78) = 2.73, p = 0.10, M included = 17.1, M excluded = 15.2), and baseline CES-D scores (F (1, 78)= 0.11, p > 0.20, M included = 26.1, M excluded = 25.2). There was, however, a significant difference between groups in regard to year in school (χ²= −0.02, p = .03, 71% freshman included, 52% freshman excluded).

Figure 1.

Study Consort.

Mobile Application

The mobile intervention was completed using the “CBT-I Coach” app. This app was originally developed the National Center for PTSD, Department of Veterans Affairs for use among veterans but is publicly available on most mobile app stores (e.g., iOS, android)^32,33. This app aims to guide users through the process of learning about sleep, developing good sleep habits, maintaining a consistent sleep/wake schedule, and limiting stimuli that can interfere with sleep. The “Learn” section of the app provides short readings that inform users about sleep mechanisms, sleep stages, and behaviors that can impact sleep. The “Tools” section of the app provides ways in which users can create healthier sleep habits. This includes setting up morning routines, reducing caffeine, muscle relaxation, and getting out of bed when they are unable to sleep (see Figure 1). The ISI and sleep diary can be found in the “My Sleep” section of the app. The ISI gives users a chance to identify their sleep difficulties and the impact these difficulties have on their life. The sleep diary allows users to track the time they got into bed, tried to fall asleep, how many times they awoke during the night, and the quality of their sleep (see Figure 2). The app provides users with a sleep efficiency score, which is the percentage of time users are asleep while in bed. Based on this score, the app generates a “sleep prescription”. This is a recommended bedtime and waketime based on the users sleep efficiency. The sleep prescription is similar to what a therapist would implement for their client when conducting sleep restriction therapy. The app also allows users to set up reminders and notifications to better aid in tracking their sleep and engaging in new sleep habits.

Figure 2.

Study Protocol.

Baseline Visit

During this initial visit on Zoom, participants were asked to read and sign the online informed consent form and complete a battery of questionnaires, which included questions regarding demographic information, Insomnia Severity Index (ISI), Epworth Sleepiness Scale (ESS), and Center for Epidemiological Studies Depression Scale (CES-D). The participants were invited to download the CBT-I Coach app on their smart phone. After the app download was completed, all participants watched a video tutorial created by the study team that taught them how to set up the app, complete sleep diaries on the app, take the biweekly ISI assessment, and find the supplemental materials within the app. A research assistant was available during the meeting to answer any questions. Upon completion of the baseline visit, participants were randomly assigned to the treatment group or a wait-list control. The study used a crossover design to evaluate both between and within subject effects of this mobile intervention on insomnia symptom. The treatment group was instructed to start using the CBT-I Coach app immediately and for the next four weeks. The participants assigned to the waitlist control group were instructed to complete the bi-weekly ISI and ESS assessments but not to start using the app until after a four-week waiting period. After the four-week waiting period, they were emailed a link to the instructional video (to reorient themselves to the app, as needed) and instructed to use the CBT-I Coach app for the subsequent four weeks (see Figure 2). All study procedures were approved by the Institutional Review Board at the University of Arkansas (protocol #2107344631)

Bi-Weekly Surveys

Every 2 weeks, all participants were asked to complete an ESS on Qualtrics and screen capture their ISI results (from the app) and upload them to an online data portal (via Qualtrics). On weeks 4 and 8 the participants were also asked to fill out measures, including the ESS and CES-D, on Qualtrics. The final survey, following the treatment period, also asked about app usage (e.g., days per week spent on the app, time per day spent on the app, feedback about the apps utility and function).

Statistical Analyses

The independent variable in this study was the intervention condition. The dependent variables of this study were ISI and CES-D scores during the week 4 and week 8 follow-ups. Means and standard deviations for each of these scores were calculated and used to estimate changes in symptoms from baseline to week four and week four to week eight. The sample was characterized by demographic information (e.g. gender, age, and education).

An apriori power analysis was conducted using G*Power 3.1 to estimate the sample size needed for the current study. Prior work using dCBT-I has shown a large effect (d = 0.92; 95% CI = 1.22–0.62) ³⁴. It is important to note that this prior work used SHUTi and not CBT-I Coach. Considering there is no clinical trials data for CBT-I Coach as a stand-alone therapy, this strategy was used to estimate a potential effect size. Using this effect size, a power of .80, and an alpha of .05, a sample size of at least 38 is needed to find significant effects. Put differently, our final sample size (n = 55) was powered to detect at least a medium effect size (d ≥ 0.68). There was attrition in the study and those participants that were lost to follow-up were not able to provide symptom data at the follow-up assessments. As a result, per protocol analyses were used to analyze all participant data. A series of hierarchical mixed-between-within-subjects ANOVAs (using the SPSS MIXED) were used to analyze the main effect of group (intervention vs. control), the main effect of time (baseline vs. week 4 vs. week 8), and the group by time interaction. Model 1 included the effects of group and time, without the interaction effect) and Model 2 included the main effects of group, time and the group by time interaction. Effect sizes were also calculated for the differences in ISI and CES-D scores among groups.

Results

Sample Characteristics

The final sample consisted of 55 college students who were predominantly women (81.8%), aged 19.9 ± 0.97 years, and white (83.6%). From this sample, 27 were randomized into the dCBT-I intervention group and 28 were randomized into the waitlist control group. No statistically significant group differences were observed at baseline for age, gender, race, employment status, baseline ISI, and baseline CES-D scores. The groups did, however, significantly differ in regard to year in school (p = 0.03), such that the intervention group included a greater proportion of Freshman (see Table 1).

Table 1.

Demographic Information

	Intervention Group (n=27)	Waitlist Control Group (n=28)	F/ χ2	P
Age in years, mean	18.85	19.82	F = 2.58	0.11
Female, count (%)	23 (85.2)	22 (78.6)	0.07	0.49
White (%)	21 (77.8)	25 (89.3)	0.01	0.32
College Freshman (%)	23 (85.2)	16 (57.1)	0.23	0.03
Unemployed (%)	19 (70.4)	18 (64.3)	0.06	0.61
ISI Baseline, mean	17.6	16.6	F = 0.85	0.36
CEDS Baseline, mean	28.4	23.9	F = 2.99	0.09

Pre-to-Post Treatment Changes in Insomnia Symptom Severity

Mixed between-within-subjects ANOVAs were used to examine the effects of the dCBT-I intervention on participants ISI scores across baseline, week 4, and week 8. The results of these analyses suggested that while there was no overall effect of group, F(1, 56) = 0.48, p > 0.20, there was an overall effect of time, F(2, 96) = 39.56, p < 0.001 (Model 1). This was qualified by a significant group by time interaction, F(2, 95) = 4.12, p = 0.02 (Model 2). Additional analyses revealed that, on average, participants in the intervention group reported a significant reduction in ISI scores from baseline to week 4, baseline M (SD) = 17.6 (3.5), week 4 M (SD) = 11.8 (4.5), Δ= −5.8, t = −5.97, p < 0.001. Participants in the waitlist group also reported a reduction in ISI scores from baseline to week 4, baseline M (SD) = 16.6 (4.6), week 4 M (SD) = 14.5 (4.9), Δ= −2.1, t = −2.23, p = 0.03. Put together, this corresponded in nearly a 4-point greater reduction in ISI scores among participants in the intervention group compared to the waitlist control group, t = 2.69, p < 0.01, Hedge’s g = 0.88. From week 4 to week 8 there was slight rebound in ISI scores for the intervention group (week 8 M [SD] = 12.6 [4.4], Δ = 0.8, t = 1.11, p > .20). However, the overall treatment effects were sustained from baseline to week 8 (Δ = −4.9, t = −6.15, p < 0.001). The downward trend seen in the waitlist group was maintained from week 4 to week 8 resulting in a mean score of 12.8 (Δ = −1.7, t = −2.23, p = 0.02). See Figure 3 and Table 2 for all model estimates.

Figure 3.

Mean ISI Scores by Group.

Table 2.

ISI Score Estimates

	Intervention Group (n)	Intervention Group ISI Scores	Waitlist Group (n)	Waitlist Group ISI Scores	p value	Hedges’ g
Baseline (SD)	40	17.6 (3.5)	40	16.6 (4.6)	-	-
Week 4 (SD)	27	11.8 (4.5)	28	14.5 (4.9)	-	-
Week 8 (SD)	24	12.6 (4.4)	22	12.8 (4.7)	-	-
Δ Baseline – Wk4 (SD)	-	−5.8 (4.4)	-	−2.1 (3.9)	0.001	0.88
Δ Wk4 – Wk8 (SD)	-	0.8 (3.8)	-	−1.7 (4.0)	0.43	0.64

Pre-to-Post Treatment Changes in Depressive Symptoms

Mixed between-within-subjects ANOVAs were used to examine the effects of the dCBT-I intervention on participants CES-D scores across baseline, week 4, and week 8 (see Figure 4). The results of these analyses suggested that there was no overall effect of group, F(1, 56) = 1.18, p > 0.20 and no overall effect of time, F(2, 98) = 0.23, p > 0.20. There was, however, a significant group by time interaction, F(2, 96) = 4.30, p = 0.02. Further analyses revealed that, on average, participants in the intervention group reported a small, non-significant reduction in CES-D scores from baseline to week 4, baseline M (SD) = 28.4 (7.1), week 4 M (SD) = 25.4 (8.8), Δ= −3.0, t = −1.57, p = 0.12. Participants in the waitlist group had a slight increase in CES-D scores from baseline to week 4, baseline M (SD) = 23.9 (11.6), week 4 M (SD) = 24.9 (14.2), Δ= 1.0, t = 0.55, p > 0.20. From week 4 to week 8 there was a rebound in CES-D scores for the intervention group (week 8 M (SD) = 29.0 (10.4), Δ = 3.6, t = 2.30, p = 0.02). The waitlist group had a reduction in CES-D scores from week 4 to week 8 resulting in a mean score of 22 (Δ = −2.9, t = −1.81, p = 0.07. See Table 3 for all model estimates.

Figure 4.

Mean CES-D Scores by Group.

Table 3.

CES-D Score Estimates

	Intervention Group (n)	Intervention Group CES-D Scores	Waitlist Group (n)	Waitlist Group CES-D Scores	p value	Hedges’ g
Baseline (SD)	40	28.4 (7.1)	40	23.9 (11.6)	-	-
Week 4 (SD)	27	25.4 (8.8)	28	24.9 (14.2)	-	-
Week 8 (SD)	24	29 (10.4)	22	22 (14.4)	-	-
Δ Baseline – Wk4 (SD)	-	−3.0 (6.3)	-	1.0 (9.2)	0.13	0.51
Δ Wk4 – Wk8 (SD)	-	3.6 (7.4)	-	−2.9 (6.4)	0.004	0.94

Feasibility and Acceptability of CBT-I Coach

Self-reported app usage was assessed at the conclusion of the study. Participants answered questions about how many weeks they used the app, on average how many days a week they engaged with the app, and if they used the tools and learning functions. Means and percentages were calculated. On average, participants reported using the app 5 days per week, with the largest percent of engagement being 7 days a week (n = 19, 37.2%). 58.8% (n = 30) of the students reported using at least one of the “Tools” (e.g. winding down) while 35.3% of the students (n = 18) reported using at least one of the “Learning” materials (e.g. Sleep 101). The students rated the overall effectiveness of the app, tools, and learning materials using a 1 (not effective) to 5 (very effective) Likert scale. The overall app effectiveness rating was 2.9 with 43% (n = 22) of the sample rating the app as moderately effective. Similarly, 46% (n = 22) of the sample rated the tools as moderately effective with a mean rating of 2.5. As for the learning material, 33% of the sample (n = 14) rated them as not effective while 31% (n = 13) of the sample rated the material as moderately effective. The mean learning effectiveness rating was 2.3 (see Figure 5).

Figure 5.

App Self-Reported Effectiveness Ratings

Discussion

In the current study, the treatment effects for CBT-I Coach were not as robust across the course of the study compared to what is observed with traditional CBT-I, however, these results do lend some support for using alternative approaches to CBT-I that are compatible with a specific population (i.e., mobile applications and college students). While the treatment effects on insomnia symptom severity remained significant for the intervention group across the duration of the study, there was a slight upward trend in ISI scores four weeks post-intervention. The self-guided nature of the application may provide easy and quick access to core skills found in CBT-I but may not be as effective at creating the long-term effects that are typically seen in conventional CBT-I ^35,36. However, these findings generally support that a self-guided, mobile intervention for insomnia may be a useful strategy for mitigating the effect of insomnia symptoms among college students, especially in situations where alternative CBT-I interventions are not available.

The group differences at post-treatment (week 4 follow-up) highlighted the moderate effectiveness of utilizing a self-guided mobile intervention to target insomnia symptoms. For example, after four weeks of using CBT-I Coach, participants reported, on average, nearly a 6-point reduction in ISI scores (33% symptom reduction). While the waitlist control group also reported a significant reduction in ISI scores, the intervention group experienced nearly a three times greater reduction in symptoms compared to the control group. The slight reduction in scores seen from baseline to week 4 among participants in the waitlist control group may have been a result of being enrolled in the study or regression to the mean. During the second 4-week period (during which the waitlist group received the CBT-I Coach intervention), the control group reported additional small, but significant, reductions in ISI scores. In contrast, the intervention group showed a small but significant rebound in insomnia symptoms 4 weeks post-intervention. It is unclear if this is a result of the length of engagement with the app or some other factor. Increasing the time spent utilizing the app (length greater than four weeks) may result in more long-lasting treatment effects after completion of the intervention. In fact, previous studies analyzed the interaction between treatment duration and insomnia outcomes. A meta-analysis conducted in 2017 showed that longer duration of digital treatment resulted in larger effect sizes³⁷.

Insomnia is a predictive and perpetuating factor of depression ^38,39. Our CES-D score analyses provide marginal support that utilizing self-guided dCBT-I may also be effective in targeting subsequent depressive symptoms along with insomnia in the short-term. More robust results have been modeled in previously conducted studies. In 2022, Kuhn et al. ²⁸ found that at follow-up after 12 weeks of utilizing an unguided mobile CBT-I intervention, participants experienced a significant reduction in depression symptoms assessed using the Patient Health Questionnaire (d = −0.8, p = .012). Similar to the ISI outcome, the intervention group experienced depression symptom reductions during the first four weeks (11% symptom reduction) with a significant rebound in scores at week eight. The waitlist group also experienced a reduction in depression symptoms while using the intervention app. However, the drop in scores was not significant. Despite our study not having more conclusive depression outcomes, it still lends support to the growing literature. Even more important, having easily accessible intervention options (self-guided dCBT-I) that address insomnia symptoms may be a strategy healthcare providers could implement to offset the elevated risk of developing depression.

The results of our analyses and participant self-reports highlight the acceptability of using an unguided dCBT-I app to treat insomnia symptoms among college students. Approximately 69% of the sample found the app to be moderately to extremely effective. Students’ comments about the app overall included, “the app was easy to use” and “the app worked as intended”. Students were also able to make suggestions about how to improve the app. Most of the recommendations involved the appearance of the app (e.g. “make the design of the app more appealing”) or the notification feature embedded with the app (e.g. “I set up the notifications to track my sleep but they never went off even when they were turned on in my settings”). That said, since the onset of this study, a new version of CBT-I Coach has been released (version 3.0) and still freely available from the app store. Further analyses showed that 46% of the students found the learning materials moderately to extremely effective and 58% found the tools to be effective. In combination with the ISI outcomes, it seems that a self-guided dCBT-I is an acceptable intervention for a college student population. Future research, however, should focus on identifying what learning materials and tools could be more effective for a student population (e.g. learning material about memory and sleep or the impact of substance use on sleep).

This study had several strengths and limitations. The first strength of this study is the ease of access to the intervention. The app is available for iOS and android users across many smart devices (e.g. smartphones or tablets) with all of the app’s features being available without internet (e.g. sleep diary). The second strength is the cost of treatment. The app is available for free across all platforms making dissemination easier for health care providers. The third strength is the app is user friendly. This could lessen the demand for CBT-I specialists or save in-person resources for those who suffer more severe or chronic forms of insomnia. Lastly, the students were instructed to use the app as often as they liked with very little to no usage rules. This increases the ecological validity of the findings. The first limitation is the study’s small sample size. A substantial number of participants were excluded from our analyses, primarily due to loss to follow-up. Increasing adherence to the treatment course and recruiting more participants would be a way to address this limitation. Another limitation of the study is utilizing a waitlist control group. Due to this study design, we were unable to assess how dCBT-I compares to active treatment (e.g., traditional CBT-I or other alternative forms of CBT-I, such as BBTI). However, the aim of this study was to take a “first look” at the possibility of using this type of intervention which was possible using this study design. An additional limitation is that, for practical reasons, all participants were introduced and oriented to the CBT-I Coach app during the baseline visit. Because of this, it is possible that participants in the control group could have started using the app during the first 4-week period. The implications of this are that the effect sizes may have been attenuated by anyone (in the control group) that started using the app prior to being instructed to. We also did not control for medication or substance use (e.g., hypnotic or cannabis use) and/or other forms of psychotherapy. This could impact our ability to draw definite conclusions that the symptom reduction experienced by the participants was due to the intervention. In an attempt to address this issue, randomization was used. Another limitation of the study was not assessing for engagement with the key components of CBT-I (e.g. sleep restriction and stimulus control). The variability in app use among students made it difficult to disconcert if a reported reduction in insomnia symptoms is due to the intervention or a natural regression to the mean. Similarly, the ambiguity of which components of CBT-I were utilized by participants may weaken the interpretability of the results. However, questions about what tools/resources the participant utilized during treatment and how often they used the app was included to the final survey. It is important to note that the self-report nature of these survey questions can also be considered a limitation of this research design. An alternative would be the “screen time” function found on smart phones and tablets. This would give future researchers more information about how long a participant is engaging with the application. Future studies would benefit from increasing their sample size, including an active control group, controlling for app engagement, and excluding or at least controlling for those participants who are utilizing medications for sleep.

Despite these limitations, these results further support the growing body of literature surrounding self-guided digital CBT-I. These findings have broader dissemination implications. More specifically, this app provides a psychoeducational tool that is adapted from an empirically supported treatment, easily accessible by the general population (due to availability through any app store), and extremely cost-effective. The implementation of dCBT-I in settings that are not properly equipped to provide face to face CBT-I may benefit from providing this type of resource to their patients. Utilization of dCBT-I can also be a strategy implemented by healthcare providers to address symptoms that are often comorbid with other mental health disorders. These results add to the growing literature that supports the use of digital CBT-I to treat insomnia ^28,29,40. Even more importantly, these findings highlight the utility of disseminating empirically based forms of treatment to populations that may be missed by healthcare providers or have other barriers to access (e.g. college students). Future research should focus on studying the effects of dCBT-I on insomnia symptoms within a more diverse sample or groups that developed insomnia following a medical diagnosis (e.g. cancer). Increasing the length of treatment may also be a factor that is addressed by future research.

Data availability statement

The data underlying this article will be shared on reasonable request to the corresponding author.