Treating comorbid insomnia in patients enrolled in therapist-assisted transdiagnostic internet-delivered cognitive behaviour therapy for anxiety and depression: A randomized controlled trial

Abstract

Transdiagnostic Internet-delivered cognitive behaviour therapy (ICBT) for patients experiencing anxiety and depression can produce large improvements in symptoms. Comorbid insomnia is common among individuals seeking treatment for anxiety and depression, yet transdiagnostic ICBT rarely targets insomnia and many ICBT patients report that symptoms of insomnia remain after treatment. This trial explored the impact of including a brief intervention for insomnia alongside an existing transdiagnostic ICBT course that included brief weekly therapist assistance. Patients were randomly assigned to receive either the Standard transdiagnostic (n = 75) or a Sleep-Enhanced course (n = 142), which included information on sleep restriction and stimulus control. Intent-to-treat analyses using generalized estimating equation (GEE) showed significant, large reductions in all primary outcomes (insomnia: d = 0.96, 95 % CI [0.68, 1.24]; depression: d = 1.04, 95 % CI [0.76, 1.33]; and anxiety: d = 1.23, 95 % CI [0.94, 1.52]) from pre-treatment to post-treatment, with changes maintained at 3-months. Patients assigned to the Sleep-Enhanced course reported larger reductions in insomnia than patients in the Standard transdiagnostic course (Cohen's d = 0.31, 95 % CI [0.034, 0.60]) at post-treatment but no significant between-group differences in any of the primary outcomes were found at follow-up. Patient-reported adherence to sleep restriction guidelines (p = .03), but not stimulus control instructions (p = .84) was associated with greater reductions in insomnia symptoms during the course. Overall, patients who received the Sleep-Enhanced course were satisfied with the materials and most patients reported making sleep behaviour changes. The trial results demonstrate that including a brief intervention targeting insomnia can be beneficial for many patients who enroll in ICBT primarily for symptoms related to anxiety and depression.

Highlights

• •Patients in both courses reported large reductions in insomnia.
• •Insomnia outcomes were better for the Sleep-Enhanced course at post-treatment.
• •Between-group differences were no longer present at 3-month follow-up.
• •Patient adherence to sleep restriction resulted in greater improvements in insomnia.

1. Introduction

Sleep difficulties such as insomnia are highly prevalent among patients with symptoms of depression or anxiety (Johnson et al., 2006), and can contribute to medical and psychiatric comorbidities (Roth et al., 2006; Stein et al., 2008). Symptoms of insomnia include difficulty falling or staying asleep, or early morning awakening with an inability to return to sleep (American Psychiatric Association, 2013). It has been estimated that 10 % of the adult population meets criteria for an insomnia disorder, with high rates of chronicity (Morin and Jarrin, 2022). Insomnia is also associated with a range of long-term personal and societal consequences such as reduced work productivity and absenteeism (Daley et al., 2009), economic burden for the healthcare system (Ozminkowski et al., 2007), and increased costs for sick leave and short- or long-term disability (Kleinman et al., 2009).

Insomnia and depression have a high comorbidity rate with estimates varying widely depending on the criteria used and the population studied. One study of 11,329 adults in the United States found that 33.6 % of patients who reported symptoms consistent with insomnia also reported symptoms consistent with depression (Hayley et al., 2014). Insomnia often precedes the onset of depressive symptoms among individuals with depression, and insomnia can persist after the successful treatment of depression (Vargas and Perlis, 2020). Findings from a meta-analysis (Baglioni et al., 2011) suggest that insomnia predicts the later development of depression, leading some researchers to describe insomnia as a risk factor for depression (e.g., Vargas and Perlis, 2020). A number of neurobiological factors are thought to mediate or moderate the relationship between insomnia and depression, including chronotype (one's preference for evening vs. morning) (Chan et al., 2014), circadian rhythm disruption (Germain and Kupfer, 2008), and gender and hormonal differences (Marver and McGlinchey, 2020). Mounting evidence suggests that insomnia predicts development of depression (Riemann and Voderholzer, 2003), highlighting the need for adequate treatment of insomnia. It has also been reported that persistent insomnia is a risk factor for future anxiety (Morphy et al., 2007).

Cognitive behaviour therapy for insomnia (CBT—I) is considered the first-line treatment for insomnia (Okajima et al., 2011) and has been found to be more effective than pharmacotherapy in the long-term (Rios et al., 2019). The two core behavioural components of CBT-I are stimulus control and sleep restriction (Edinger and Carney, 2014). Stimulus control emphasizes restricting the bedroom to sleep and sex to remove the behavioural association between the bedroom and high arousal states that can be elicited through eating, working, or watching television (Edinger and Carney, 2014). The goal of sleep restriction is to increase sleep efficiency by creating a new “sleep window” (i.e., amount of time spent in bed) and then gradually titrating the amount of time in bed contingent upon increased sleep efficiency (Miller et al., 2014). Sleep restriction can help to restore the patient's circadian rhythm and increase the ease of falling asleep (Maurer et al., 2018).

CBT-I is commonly offered in a disorder-specific format to patients with insomnia. Although the core components of CBT-I are stimulus control and sleep restriction, CBT-I programs often include treatment content and components common to CBT for depression and anxiety, such as psychoeducation, cognitive restructuring, and relaxation training. Given the overlap in approach between CBT for insomnia, depression, and anxiety, an integrative or transdiagnostic approach to treating these mental health concerns may offer benefits to both the patient and therapist. Transdiagnostic approaches to therapy involve the identification of psychological processes that are relevant across a range of mental disorders, which can then be targeted with interventions that apply to a broad range of concerns (Sauer-Zavala et al., 2017). A major benefit of a transdiagnostic approach for patients is that they can address comorbid concerns within a single course of therapy, minimizing repetition of material and potentially result in better understanding of how their concerns relate to one another (Barlow et al., 2017). Transdiagnostic approaches benefit therapists and program administrators in that they reduce the need for training in multiple disorder-specific protocols (Păsărelu et al., 2017).

Internet-delivered CBT (ICBT) is an effective alternative to face-to-face CBT (Andersson et al., 2019; Etzelmueller et al., 2020; Karyotaki et al., 2021) that can reduce patient barriers to accessing evidence-based care (Andersson, 2016). A recent meta-analysis found that both therapist-assisted and self-guided ICBT for insomnia produced medium effect sizes for improvements in insomnia (Simon et al., 2023), and that longer treatment duration and more therapist support were related to improved outcomes (Zachariae et al., 2016). An earlier meta-analysis of ICBT for insomnia found that therapist-assisted ICBT outperformed self-guided programs (Hasan et al., 2022) and studies on ICBT for depression suggest that therapist assistance can be beneficial for patient outcomes, especially among clients with clinical levels of depression (Karyotaki et al., 2021). Therefore, therapist assistance is likely of benefit in a transdiagnostic ICBT program for patients experiencing symptoms of depression and/or anxiety with comorbid insomnia symptoms. However, there are no known studies in which key components of CBT-I have been included and evaluated within transdiagnostic ICBT for depression and anxiety. The current trial is the first to specifically investigate whether simultaneously treating insomnia among patients enrolled in transdiagnostic ICBT for anxiety and depression can offer patients relief from symptoms of insomnia without reducing the effectiveness of the program for symptoms of anxiety and depression. The potential to treat insomnia within a transdiagnostic program offers the chance for efficient service delivery that could be most impactful for clinics that do not have resources to offer many different online treatment programs.

The current trial investigated the following: 1) whether including components of CBT-I as part of a transdiagnostic ICBT course for anxiety and depression can produce further reductions in symptoms of insomnia, anxiety, and depression; 2) whether patient reports of the frequency of engagement in certain sleep behaviours (i.e., sleep restriction and stimulus control) are correlated with the degree of improvement in insomnia symptoms; 3) patient satisfaction and acceptability of the inclusion of CBT-I materials; and 4) patient challenges and opportunities to improve the CBT-I materials. It was hypothesized that patients who received the additional CBT-I material would report larger reductions in symptoms of insomnia compared to patients who received the standard transdiagnostic ICBT course (Hertenstein et al., 2022). Further, it was hypothesized that more frequent engagement in sleep restriction and stimulus control would be positively correlated with patient-reported change in insomnia symptoms. No specific hypotheses were made regarding patient satisfaction with the materials, as this was considered an exploratory analysis within the trial.

2. Method

2.1. Design, power, and ethics

The current study employed a two-group randomized controlled trial design, with outcomes assessed at post-treatment and three-month follow-up. Patients who sought ICBT for symptoms of anxiety and depression who also displayed symptoms of comorbid insomnia were randomly assigned (using a 1:2 ratio) to either receive the Standard transdiagnostic program, called the Wellbeing Course (control), or to receive the Sleep-Enhanced Wellbeing Course. Consistent with the framework provided by Goldberg et al. (2023) on the selection of control conditions for trials of mobile health interventions, this trial used an additive design. It is notable that the control condition was an active, treatment-as-usual intervention which included an existing brief psychoeducational resource with sleep hygiene tips. A meta-analysis of ICBT programs for insomnia found an average hedge's g effect size of 0.49 for reductions in insomnia and improvements in subjective sleep quality (Zachariae et al., 2016). Using this effect size, a power analysis was conducted with 0.8 power and a 0.05 significance level, and it was determined that each treatment group would require 66 patients. The sample size for the Sleep-Enhanced program was doubled to increase the power to detect the association between reduction in insomnia and adherence to sleep restriction and stimulus control. Therefore, the target total sample size was 198 patients and we planned to admit all individuals to the trial who applied during a predefined admission period which was expected to exceed this requirement.

Ethics approval was received through the University of Regina Research Ethics Board and the trial was registered through ClinicalTrials.gov (#NCT04512768). Two changes were made to the trial design after trial registration. Specifically, the Working Alliance Inventory and Sheehan Disability Scale were not administered to patients due to feedback during the pilot study that the questionnaires were too long.

2.2. Patient recruitment, screening, and randomization

The current study was conducted within a routine-care ICBT clinic in Canada using standard recruitment procedures, which are described in detail elsewhere (e.g., Hadjistavropoulos et al., 2022a, Hadjistavropoulos et al., 2022b). In brief, patients who completed the online screening for the Wellbeing Course during the trial period (September 9, 2020 to December 29, 2020) completed the Insomnia Severity Index (ISI; Morin et al., 2011) to assess for presence of insomnia symptoms. Patients who endorsed significant symptoms of insomnia (i.e., ISI ≥ 10) were randomized to the Sleep-Enhanced Wellbeing Course or the Standard Wellbeing Course. Patients who did not endorse significant symptoms of insomnia were not included in the current trial and instead automatically received the Standard Wellbeing Course.

Prospective patients had to meet the following inclusion criteria for the trial: be at least 18 years of age; endorse at least mild symptoms of depression and anxiety, with a score ≥ 5 on the Patient Health Questionnaire 9-item (PHQ-9; Kroenke et al., 2001) or Generalized Anxiety Disorder 7-item (GAD-7; Spitzer et al., 2006); endorse significant symptoms of insomnia (ISI ≥ 10); reside in Saskatchewan for the duration of the 8-week program; have access to a computer and Internet; and provide an emergency medical contact. Reasons for exclusion from the trial included: primary concerns of unmanaged mania, psychosis or substance abuse; high risk for suicide (based on intent/planning assessed in a clinical interview); hospitalization for mental health in the past year; receiving mental health treatment more than twice a month; other sleep concerns (i.e., sleep apnea, restless leg syndrome, extreme nightmares, and shift work), or other health concerns that would interfere with patients' ability to complete ICBT. Patient flow is outlined in further detail in Fig. 1.

Fig. 1.

Flow chart of patient randomization and completion rates.

Patients who were deemed eligible for the trial were randomly assigned at the conclusion of the screening process. A computer-generated randomization function that is available as part of the REDCap study management software package (REDCAP, n.d.) was used to assign the patients to one of the two interventions on a 1:2 ratio as explained above.

2.3. Interventions

2.3.1. Standard Wellbeing Course

The Standard Wellbeing Course is an 8-week ICBT course developed at Macquarie University (Dear et al., 2015; Titov et al., 2015) that consists of five lessons. In lesson 1, patients are introduced to the CBT model and receive psychoeducation about symptoms of depression and anxiety. Lesson 2 focuses on unhelpful thinking patterns and introduces thought challenging as a strategy for managing these patterns. Lesson 3 focuses on physiological over- and under-arousal and includes controlled breathing and activity planning to manage these symptoms. In lesson 4, patients learn about avoidance and safety behaviours and are introduced to graded exposure. The focus of lesson 5 is on relapse prevention and goal setting. Patients can also access several additional resources throughout the course (e.g., assertive communication, managing worry, structured problem solving, managing panic), including a brief resource on sleep hygiene strategies. The resource on sleep hygiene strategies does not include information about sleep restriction or stimulus control.

2.3.2. Sleep-Enhanced Wellbeing Course

The Sleep-Enhanced Wellbeing Course includes an additional lesson developed specifically for sleep based on content from CBT—I, with an emphasis on sleep restriction and stimulus control. The lesson was offered at the beginning of treatment, before the core Wellbeing Course lessons, to maximize the time that therapists can support patients with sleep restriction and stimulus control. Patients had to complete the sleep lesson before beginning lesson 1 of the Wellbeing Course. There was no delay before accessing lesson 1, so the total duration of the course was the same as the Standard Wellbeing Course (8 weeks).

Materials for the new lesson were developed in consultation with the patient-oriented research steering committee (PORSC), which included patient partners with lived experience, clinicians, and health care decision-makers. Materials underwent four rounds of revision based on feedback from the PORSC before a brief pilot study (n = 25) was conducted to gather patient feedback. The materials were reviewed and revised an additional two times before use in the current trial. The sleep lesson consisted of 43 slides and a downloadable 13-page ‘Do-it-Yourself’ guide. The final sleep materials included: 1) psychoeducational components (i.e., about the circadian rhythm; the relationship between anxiety, depression, and insomnia; and the impact of unhelpful beliefs about sleep and insomnia) and 2) behavioural components (i.e., stimulus control and sleep restriction). The five slides on stimulus control provided a definition of stimulus-response relationships and highlighted how people with insomnia often have an unhelpful stimulus-response relationship with bed (e.g., associating the bed with worrying or alertness). Information was provided about developing a new stimulus-response relationship, minimizing arousing activities in bed, and getting out of bed after 15–20 min to do non-arousing activities if unable to fall asleep. The slides also included a definition of sleep restriction and instructions for patients to: 1) track their sleep for one week; 2) plan to stay in bed only for the amount of time they slept in the previous week; and 3) track time in bed awake versus time sleeping and adjust their sleep schedule using a sleep restriction decision guide provided in the slides. In addition to the core sleep resource (presented to patients at the beginning of the course), a brief callback to the topic of insomnia was added after each of the core Wellbeing Course lessons. These supplementary slides were designed to help patients apply the lesson to their difficulties with sleep (e.g., after Lesson 2 on thought challenging, a brief discussion of how to apply thought challenging to beliefs about sleep was added). Additionally, patients completed a weekly questionnaire about sleep that was also designed to serve as a reminder to engage in sleep restriction and stimulus control behaviours.

2.4. Measures

All prospective patients completed an online screening questionnaire to assess for eligibility. If initial inclusion criteria were met, they were asked to provide demographic information (i.e., age, gender, education, employment and relationship status, and whether they lived in a city or rural location). Patients completed measures of symptoms of depression, anxiety, insomnia, panic, and social anxiety at pre-treatment, post-treatment, and 20-week follow-up (described below). Additionally, patients completed weekly measures of depression, anxiety, and insomnia during the course.

2.4.1. Primary measures

2.4.1.1. ISI (Morin et al., 2011)

The ISI is a 7-item self-report measure of symptoms of insomnia. Respondents answer questions about difficulties falling asleep, staying asleep, and waking too early, as well as the impact of these symptoms on functioning and distress. Total scores on the ISI range from 0 to 28, with higher scores indicating more severe symptoms of insomnia. Scores ≥10 have been used to detect cases of insomnia (Morin et al., 2011).

2.4.1.2. PHQ-9 (Kroenke et al., 2001)

The PHQ-9 is a 9-item self-report measure of symptoms of depression. Total scores range from 0 to 27, with a cut-off score of 10 or above used as an indicator for major depressive disorder (Manea et al., 2012), and a six-point change indicating a reliable change in symptoms (Gyani et al., 2013).

2.4.1.3. GAD-7 (Spitzer et al., 2006)

The GAD-7 includes 7 self-report items to assess symptoms of generalized anxiety. Total scores range between 0 and 21, with a score of 10 or above used as an indicator for generalized anxiety disorder (Spitzer et al., 2006), and a 4-point change indicating reliable change in symptoms (Gyani et al., 2013).

2.4.2. Secondary measures

2.4.2.1. Sleep diary

Patients completed a 7-item questionnaire about their sleep on a weekly basis. Patients were asked to rate their sleep quality on a scale ranging from 1 (Extremely Poor Sleep) to 7 (Excellent Sleep Quality) and their fatigue on a scale from 1 (No Fatigue) to 7 (Incapacitating Fatigue). They were also asked to estimate the number of hours they spent in bed each night during the week, as well as how many of the hours in bed they spent awake (0–16 in 0.25-hour increments). Finally, patients indicated how many days in the past week they engaged in each of the following: sleep restriction, stimulus control, and napping. Data from the sleep diary was used by clinicians to monitor patient progress and provide feedback. Sleep diary data was used in analysis to examine factors that affected patients' sleep throughout the course.

2.4.2.2. Panic Disorder Severity Scale – Self Report (PDSS-R; Houck et al., 2002)

The PDSS-SR includes 7 self-report items related to symptoms of panic disorder. Total score range between 0 and 28 and a cut-off score of 8 has been used as an indicator for the diagnosis of panic disorder (Allen et al., 2016).

2.4.2.3. Social Interaction Anxiety Scale and Social Phobia Scale – Short Form (SIAS-6/SPS-6; Peters et al., 2012)

The SIAS-6/SPS-6 includes 12 self-report items related to symptoms of social anxiety. Cutoff scores of 7 and above on the SIAS-6, or 2 and above on the SPS-6 appear optimal for detecting individuals with social anxiety disorder (Peters et al., 2012). Scores on the SIAS-6 and SPS-6 can be summed to create a single reliable and valid measure of social anxiety (Peters et al., 2012).

2.4.2.4. Treatment satisfaction

At post-treatment, patients answered questions about how satisfied they were with treatment overall, as well as how satisfied they were with the ICBT website on a 5-point scale ranging from 1 (very dissatisfied) to 5 (very satisfied). Patients also responded to questions about whether they would feel confident recommending the treatment to a friend and if it was worth their time to do the course (‘yes’ or ‘no’).

2.5. Therapist support

Patients were contacted weekly by their therapist during the 8-week program, primarily through a secure messaging system built into the clinic software platform. Therapists followed the ICBT clinic's guidelines for what to include in messages (e.g., encouragement, direction about the course, psychoeducation, answering questions; see Hadjistavropoulos et al., 2018 for additional details). Therapists were directed to spend the same amount of time with patients in each condition (i.e., 15–20 min per patient, per week). In the Sleep-Enhanced Course, the one modification was that therapists would include brief passages in their messages to encourage patients to engage in sleep restriction and stimulus control. Each week, after patients completed lesson 1, therapists also reviewed patients' sleep diaries and offered to answer questions about sleep restriction (e.g., target sleep window calculation) or assist patients in trouble-shooting barriers. Therapists were also encouraged to help patients relate the skills in each of the core lessons to sleep. For example, therapists encouraged the use of relaxation techniques at bedtime and discussed exercise and arousal levels during the lesson focused on controlled breathing and activity planning. Therapists contacted patients by telephone if the patient: 1) had an increase of five or more points on weekly measures of depression or anxiety; 2) reported suicidality; 3) had not logged in for the last week; or 4) requested to speak with the therapist by telephone.

2.6. Analyses

Patient demographics were collected, and descriptive statistics were calculated to describe the sample. Independent t-tests were used to compare demographics between the two groups. Consistent with previous research (e.g., Hadjistavropoulos et al., 2016; Owens et al., 2019), Generalized Estimating Equation (GEE) models were employed to examine changes in symptom measures over time (from pre-treatment to the 3-month follow-up). GEE is a common longitudinal technique that emphasizes modeling changes over time while also accounting for within-subject variance through a working correlation model (Hubbard et al., 2010; Liang and Zeger, 1986). For all GEE analyses, an unstructured working correlation and maximum likelihood estimation were utilized, along with robust error estimation. To compare outcomes between the Standard and Sleep-Enhanced courses (group), the marginal models included time, group, and their interaction (time × group). A significant interaction effect indicates significant differences between the groups over time. The estimated marginal means from the GEE analyses were used to calculate the average percentage change across time. Cohen's d effect sizes and associated 95 % confidence interval were calculated for each of the outcome measures using the estimated marginal means from the GEE analysis and pooled standard deviation. Pairwise comparisons used a Bonferroni correction to adjust for multiple comparisons.

To examine whether the frequency of use of sleep restriction and stimulus control was significantly correlated with the magnitude of reductions on the ISI at post-treatment, the average number of days per week the patient reported engaging in sleep restriction and stimulus control was calculated.

Patient responses to the treatment satisfaction questionnaire were analyzed using an ANOVA to assess for differences in treatment satisfaction between the treatment conditions, with a significance level of 0.01 as a partial control for multiple comparisons. Completion rates for each lesson of the course for patients who received the Sleep-Enhanced and Standard Wellbeing Course were tabulated and chi-square tests were used to assess for differences in the proportions of patients who completed all five lessons between the two courses.

Patient responses to an open-ended question eliciting feedback about the sleep materials were analyzed using an open qualitative content analysis approach (Hsieh and Shannon, 2005). During the coding process, a research assistant (AW) experienced with qualitative content analysis in ICBT research conducted the first round of coding and then continued until saturation was reached (defined as finding no new themes in ten patient responses). The primary investigator (ME) then recoded the responses using the identified themes as a quality assurance check and to ensure no important themes had been missed. Patient responses to a question about examples of how a skill or strategy from the treatment made a difference in their life were coded using the same approach.

3. Results

3.1. Patient recruitment

Patients were recruited to the trial between September 9, 2020, and December 17, 2020, during which time 607 individuals completed the online screening for the Wellbeing Course. See Fig. 1 for details on randomization and patient flow throughout the stages of screening, accessing the course, and completing post-treatment and 20-week follow-up measures in each group.

3.2. Sample characteristics

Demographic data collected for patients in the two groups (n = 217) is displayed in Table 1. Patients were predominantly White (89.5 %; n = 194) and female (75.6 %; n = 64). Approximately half of the patients (53.9 %; n = 117) reported employment while some (20.3 %; n = 44) reported that home-based family care best described their work status. Approximately half (50.7 %; n = 110) of patients lived in a large city, with others lived in a rural area or small town (20.3 %; n = 44), or a medium town or small city (29.0 %; n = 63). Most patients reported at least some post-secondary education beyond high school (77.4 %; n = 68) and only a few (1.8 %, n = 4) reported not graduating high school. Many patients (63.6 %, n = 138) reported learning about the program from a health care professional (e.g., a doctor or other mental health professional), while others learned about the program from a friend or family (17.5 %, n = 38) or online sources (8.8 %, n = 19). Results of independent t-tests showed that patient demographic characteristics and pre-treatment symptom scores (i.e., ISI, PHQ-9, GAD-7, PDSS, and SIAS-6/SPS-6 scores) did not differ significantly between the Sleep-Enhanced and Standard Wellbeing groups.

Table 1.

Demographic characteristics of sample.

Variable	Standard wellbeing (n = 75)	Sleep-enhanced (n = 142)	All patients (n = 217)
Age, Mean (SD)	38.49 (12.90)	37.13 (12.77)	37.60 (12.80)
Range	20–67	18–68	18–68
Sex
Female	54 (72.0 %)	110 (77.5 %)	164 (75.6 %)
Male	21 (28.0 %)	31 (21.8 %)	52 (24.0 %)
Transgender	0 (0 %)	1 (0.7 %)	1 (0.5 %)
Ethnicity
White	66 (88.0 %)	128 (90.1 %)	194 (89.4 %)
Indigenous	3 (4.0 %)	6 (4.2 %)	9 (4.1 %)
Asian	3 (4.0 %)	1 (0.7 %)	4 (1.8 %)
South Asian	0 (0.0 %)	4 (2.8 %)	4 (1.8 %)
Other/prefer not to answer	3 (4.0 %)	3 (2.1 %)	6 (2.8 %)
Employment status
Paid work	44 (58.7 %)	73 (51.4 %)	117 (53.9 %)
Homemaker and/or childcare	10 (13.3 %)	34 (23.9 %)	44 (20.3 %)
Retired	4 (5.3 %)	9 (6.3 %)	13 (6.0 %)
Student	3 (4.0 %)	3 (2.1 %)	6 (2.8 %)
Unfit for work due to health problems	6 (8.0 %)	15 (10.6 %)	21 (9.7 %)
Unemployed for other reasons (e.g., involuntary unemployment or volunteer work)	8 (10.7 %)	8 (5.6 %)	16 (7.4 %)
Location
Rural area or small town (<7000 citizens)	15 (20.0 %)	29 (20.4 %)	44 (20.3 %)
Medium town/small city (7000–200,000 citizens)	29 (38.7 %)	34 (23.9 %)	63 (29.0 %)
Large city (>200,000 citizens)	31 (41.3 %)	79 (55.6 %)	110 (50.7 %)
Education
Less than high school	0 (0.0 %)	4 (2.8 %)	4 (1.8 %)
High school diploma	13 (17.3 %)	32 (22.5 %)	45 (20.7 %)
Any post-secondary education	62 (82.7 %)	106 (74.6 %)	168 (77.4 %)
Referral source
Physician or other medical professional	50 (66.7 %)	88 (62.0 %)	138 (63.6 %)
Friend or family	10 (13.3 %)	28 (19.7 %)	38 (17.5 %)
Online source	9 (12.0 %)	10 (7.0 %)	19 (8.8 %)
Employer, union, or professional association	2 (2.7 %)	5 (3.5 %)	7 (3.2 %)
Printed poster	2 (2.7 %)	0 (0.0 %)	2 (0.9 %)
Other	2 (2.7 %)	11 (7.7 %)	13 (6.0 %)

3.3. Completion rate and missing data

More than half the patients (n = 125; 57.6 %) accessed the final lesson in the 8-week course, with the proportion of patients who accessed the final lesson being similar for the Sleep-Enhanced (n = 79; 55.6 %) and Standard Wellbeing (n = 46; 61.3 %) groups (χ ²₍₁₎ = 0.65; p = .42). There was substantial patient attrition which led to missing data at both the post-treatment (35–37 %) and 20-week follow-up (39–40 %).

The analysis of missingness with Little's Missing Completely at Random (MCAR) test (χ² = 138.02, df = 96, p = .003) suggested that the data were not MCAR (Little and Rubin, 2002). Following the intention-to-treat principle and consistent with previous research, missing data on primary and secondary outcome measures were imputed using the multiple imputation method, generating 30 multiply imputed datasets so that the data from all eligible patients were included in the analysis even though data were found not to be MCAR (Enders, 2010; Graham et al., 2007). Analyses were then conducted to identify potential causes and correlates of missingness. These analyses indicated that demographic variables and pre-treatment clinical variables were not significant predictors of missing at post-treatment for primary outcomes in a binary logistic regression analysis (except for age). Furthermore, age was not a significant predictor of any of the primary outcomes at post-treatment (p range: 0.36–0.97) in linear regression analyses. Therefore, age was not included in the final GEE analysis.

3.4. Symptom measures

Table 2 presents the mean symptom scores at pre-treatment, post-treatment, and 20-week follow-up, as well as percentage changes from primary and secondary outcome measures.

Table 2.

Mean scores, percentage changes, and within-group effect sizes for primary and secondary outcomes.

								Change from pre-treatment		Within-group effect size from pre-treatment
Symptom domain		Pre-treatment		Post-treatment		20-week follow-up		To Post-treatment	To Follow-up	To Post-treatment	To Follow-up
(Measure)	n	M	SD	M	SD	M	SD	%	%	Cohen's d [95 % CI]	Cohen's d [95 % CI]
Insomnia (i.e., ISI)
Standard Wellbeing	75	15.81	4.04	11.55	6.07	9.32	7.75	26.95	41.04	0.83 [0.36–1.30]	1.10 [0.57–1.53]
Sleep-Enhanced	142	16.37	4.55	9.54	6.52	8.97	8.26	41.74	45.24	1.22 [0.857–1.57]	1.10 [0.76–1.46]
All patients	217	16.09	4.44	10.55	6.87	9.15	9.37	34.47	43.17	0.96 [0.68–1.24]	0.95 [0.67–1.23]
Depression (i.e., PHQ-9)
Standard Wellbeing	75	13.87	5.72	7.38	4.93	6.20	6.54	46.77	55.27	1.22 [0.72–1.71]	1.25 [0.75–1.74]
Sleep-Enhanced	142	14.20	5.37	6.35	5.82	7.11	7.04	55.32	49.97	1.40 [1.04–1.77]	1.13 [0.78–1.49]
All patients	217	14.04	5.89	6.86	5.77	6.65	8.13	51.10	52.59	1.23 [0.94–1.52]	1.04 [0.76–1.32]
Anxiety (i.e., GAD-7)
Standard Wellbeing	75	12.59	4.92	7.30	5.33	5.68	5.75	42.02	54.88	1.03 [0.55–1.51]	1.29 [0.794–1.789]
Sleep-Enhanced	142	13.17	4.89	6.94	5.33	6.18	6.17	47.30	53.08	1.22 [0.86–1.58]	1.26 [0.91–1.62]
All patients	217	12.88	5.16	7.12	5.85	5.93	6.69	44.72	53.96	1.04 [0.76–1.33]	1.16 [0.88–1.45]
Panic (i.e., PDSS-SR)
Standard Wellbeing	75	7.76	6.02	4.88	4.91	3.50	5.00	37.11	54.84	0.52 [0.06–0.98]	0.77 [0.30–1.24]
Sleep-Enhanced	142	8.49	5.76	4.82	5.68	3.75	5.00	43.26	55.84	0.64 [0.30–0.98]	0.88 [0.53–1.22]
All patients	217	8.13	6.23	4.85	5.76	3.63	5.89	40.32	55.36	0.55 [0.28–0.82]	0.74 [0.47–1.02]
Social anxiety (i.e., SIAS-6/SPS-6)
Standard Wellbeing	75	13.51	10.22	11.23	8.60	11.30	11.91	16.86	16.33	0.24 [−0.21–0.70]	0.20 [−0.25–0.65]
Sleep-Enhanced	142	14.65	11.66	11.31	9.55	11.52	13.02	22.81	21.42	0.31 [−0.02–0.64]	0.25 [−0.08–0.58]
All patients	217	14.08	11.34	11.27	9.62	11.41	14.15	19.95	18.98	0.27 [0.0–0.53]	0.21 [−0.06–0.47]

Note. ISI = Insomnia Severity Index; PHQ-9 = Patient Health Questionnaire-9; GAD-7 = Generalized Anxiety Disorder-7; PDSS-SR = Panic Disorder Severity Scale; SIAS-6/SPS-6 = Social Interaction Anxiety Scale/Social Phobia Scale.

3.4.1. Primary outcomes

Intent-to-treat analyses using Generalized Estimating Equation (GEE) showed a significant reduction in all primary outcomes over time (time effects for the ISI (β1 = −1.18, 95 % CI = [− 1.96, − 0.40], p = .003), GAD-7 (β1 = −1.27, 95 % CI = [− 1.90, − 0.64], p < .001), and PHQ-9 (β1 = −1.57, 95 % CI = [− 2.23, − 0.91], p < .001). A large within-group Cohen's effect size of d = 0.95, 95 % CI [0.67, 1.23] was found for insomnia. Likewise, large effect sizes of d = 1.16, 95 % CI [0.88, 1.45] and d = 1.04, 95 % CI [0.76, 1.32] were found for anxiety and depression, respectively. There were no statistically significant main effects for group (p range: 0.30–0.79) or time by group interactions (p range: 0.34–0.98) on the primary outcome measures. For all the primary outcomes, pairwise comparisons by group revealed no significant differences in improvements in scores from baseline to post-treatment (p range: 0.34–0.98) and from baseline to 3- month follow-up (p range: 0.09–0.98) except for ISI. A small but significant between-group difference was found in favour of the Sleep-Enhanced course for reductions in ISI from pre-treatment to post-treatment (Cohen's d = 0.31, 95 % CI [0.034, 0.60], p < .05). However, this difference was not sustained at follow up (p = .17).

3.4.2. Secondary outcomes

The GEE analyses revealed statistically significant time effects for the PDSS (β1 = −0.75, 95 % CI = [− 1.34, − 0.15], p = .01), whereas there was no significant effect of time on SIAS-6/SPS-6 (β1 = −0.22, 95 % CI = [− 1.24, 0.80], p = .67). There were no statistically significant main effects for group (p range: 0.49–0.54) or time by group interactions (p range: 0.59–0.64) for these measures. Effect sizes for changes in PDSS and SIAS-6/SPS-6 scores are summarized in Table 2.

3.5. Sleep behaviour analyses

The majority of patients admitted to the Sleep-Enhanced program completed at least one weekly sleep diary (90 %; n = 128), which were administered from the second week of the program onward. On average, patients completed 4.94 (SD = 2.01) out of the seven sleep diaries. Patients reported an average of 3.24 (SD = 1.87) days of sleep restriction, 4.32 (SD = 1.87) days of stimulus control, and 1.09 (SD = 1.33) naps per week.

A hierarchical regression model was evaluated for predicting post-treatment ISI scores based on pre-treatment ISI score (step 1) and the average number of days of sleep restriction and stimulus control reported by patients over the treatment period (step 2). The number of days of sleep restriction was a significant predictor of reduced ISI scores (β = −0.705, p = .03), but the number of days of stimulus control was not (β = −0.07, p = .84). The full model explained 18.2 % of the variance in ISI change scores (F_(2,88) = 6.521; p < .001). Inclusion of sleep restriction and stimulus control parameters contributed about 6.5 % of this explanatory power (R² change).

3.6. Treatment satisfaction and feedback on sleep lesson

Regardless of group, patients reported high levels of treatment satisfaction overall (M = 4.06, SD = 0.86) with no statistically significant differences on any of the treatment satisfaction questions. Most patients who completed the course reported being at least satisfied (77.4 %; n = 106). Nearly all patients reported that they thought the course was worth their time (96.4 %, n = 132) and that they would recommend it to a friend (97.1 %; n = 133).

Most patients who completed the feedback questionnaire reported being at least somewhat satisfied with the sleep materials (M = 5.45, SD = 1.58), that they would recommend the sleep materials to a friend (M = 5.80, SD = 1.52), that the materials were easy to understand (M = 6.20, SD = 1.12), and that the sleep materials were helpful at addressing their concerns (M = 5.29, SD = 1.77). Most patients also indicated that their knowledge of sleep had increased (M = 5.10, SD = 1.86) and that their sleep-related behaviours had changed (M = 4.42, SD = 1.86).

3.7. Qualitative analysis of patient feedback

Saturation was reached after analyzing 59 responses. Half the patients chose to leave the open feedback response box blank (n = 30, 50.8 %). Many responses (n = 15; 25.4 %) provided generally positive statements. Eleven responses (18.6 %) explicitly stated that the sleep information provided was important or useful. Among those who offered suggestions for improvement or criticism of the resource, 5 patients (8.5 %) indicated the materials were not relevant to their personal situation. 5 patients (8.5 %) felt the instructions provided were unclear or had trouble applying them, and 4 patients (6.8 %) reported that the information provided was not new to them. Patients reported factors that interfered with applying the suggestions to improve their sleep, such as difficulties managing sleep while caring for young children, difficulties sleeping with a partner who has their own sleep concerns (e.g., restlessness, nightmares, shiftwork), disruption of sleep schedule due to holidays/travel, late nights or other responsibilities, pets that cause nighttime waking, health issues (e.g., pain, headaches, enuresis, hormone issues), and medication side-effects. Notably, when patients were asked to provide an example of how a skill or strategy from the Wellbeing Course made a difference in their life, several patients (n = 9; 12.5 %) included sleep skills in their response.

4. Discussion

4.1. Insomnia symptoms

This trial is the first to investigate whether simultaneously treating insomnia among patients who sought out and enrolled in transdiagnostic ICBT for symptoms of depression and anxiety can improve patient outcomes. Based on previous results (Hertenstein et al., 2022), the Standard Wellbeing Course was expected to produce a moderate reduction in insomnia symptoms, and indeed a 4.26 point (26.95 %) reduction in ISI scores was reported by patients assigned to the Standard Wellbeing course between pre-treatment and post-treatment. Patients randomly assigned to the Sleep-Enhanced condition reported a larger mean ISI reduction of 6.89 points (41.74 %) between pre-treatment and post-treatment; however, both groups reported similar reductions in symptoms at 20-week follow-up. A meta-analysis of 15 randomized control trials comparing ICBT for insomnia to wait-list controls found that ICBT resulted in a decrease in ISI scores from pre- to post-treatment of 4.3 points (95 % CI: −7.1, −1.5; p = .017) more than wait-list control (Seyffert et al., 2016). In the present trial, patients in the Sleep-Enhanced condition reported a change between pre-treatment and post-treatment that was 2.57 points greater than those in the Standard condition, which notably was an active treatment condition (as opposed to a wait-list control) that included basic sleep hygiene advice. The degree of severity reduction observed in the current trial based on past research findings corresponds to a slight clinical global improvement (Morin et al., 2011), solidifying support for targeting insomnia as part of ICBT for anxiety and depression. It also further supports the finding that methods in CBT-I can be effective when co-morbid symptoms are present (Hertenstein et al., 2022). Nevertheless, it is important to note that by 20-week follow-up, both the Standard Wellbeing Course and Sleep-Enhanced Course had similar large effects for insomnia. It does appear that the Standard Wellbeing Course alone positively impacts insomnia symptoms, which was also found in a subsequent observational study of the Wellbeing Course (Peynenburg et al., 2022). It is possible that patients with insomnia symptoms make use of general cognitive strategies from the Wellbeing Course, such as thought challenging, to target unhelpful beliefs about sleep. Although CBT-I interventions typically emphasize behavioural strategies such as sleep restriction and stimulus control, Sunhed et al. (2020) found that internet-delivered behaviour therapy and internet-delivered cognitive therapy produced equivalent improvements in insomnia symptoms. These findings may help to explain the equivalent outcomes between the Standard and Sleep-Enhanced Course that were found at 20-week follow-up.

The results of the present trial therefore suggest that including elements of CBT-I in a transdiagnostic course may lead to faster improvement in insomnia, but ultimately similar long-term outcomes. Future research could seek to replicate the present trial but measure symptoms more frequently and over longer follow-up periods to further clarify rates of symptom change and long-term outcomes.

4.2. Depression and anxiety symptoms

Consistent with previous ICBT research (e.g., Hadjistavropoulos et al., 2021), large reductions in symptoms of anxiety and depression were observed at post-treatment and maintained at 20-week follow-up. The results are comparable to a meta-analysis of 19 other ICBT trials for anxiety and depression including a total of 2952 patients, which demonstrated medium to large effect sizes for anxiety and depression (Păsărelu et al., 2017). The results suggest that inclusion of additional information on insomnia did not negatively impact anxiety and depression outcomes by diluting attention to core strategies in the course represent an important contribution.

4.3. Sleep-related behaviours

Patients who reported adhering to sleep restriction guidelines reported greater reductions in insomnia symptoms at post-treatment, while engagement in stimulus control was not related to insomnia symptoms at post-treatment. This is consistent with previous research (Maurer et al., 2022) which has found that sleep restriction is a particularly valuable skill for patients with insomnia symptoms and comorbid anxiety and depression. Engaging in stimulus control may still be a useful strategy, but the current trial may have been insufficiently powered to detect an effect, or patients may have already learned parts of stimulus control through sleep hygiene advice that was included in the Standard Wellbeing materials. A recent meta-analysis finds that stimulus control can be helpful in isolation for participants, but questions its mechanisms of action (Verreault et al., 2023). In the current study, most patients reported their sleep behaviours had at least somewhat changed. Patients reporting no changes in their sleep behaviours commonly reported already being familiar with the behavioural strategies presented. The benefits of engaging with the sleep intervention, and sleep restriction in particular, were consistent with previous ICBT for insomnia research results (Kraepelien et al., 2021). Future research should focus on ways to improve patient engagement with sleep restriction and stimulus control to ensure patients can experience greater benefit. For example, a previous study found that including seven 15-minute telephone calls with a therapist significantly improved the therapeutic effects of bibliotherapy CBT—I, compared to no therapist support (Kaldo et al., 2015). These findings suggest that the inclusion of brief telephone calls can help increase patients' engagement with sleep behaviour strategies.

4.4. Patient acceptability

High ratings of overall satisfaction, understandability, helpfulness, and increased knowledge, combined with most patients reporting that they changed their sleep-related behaviours suggest that patient acceptability for the new sleep materials was high. The finding that patient acceptability was high in the current trial is consistent with findings of a recent network meta-analysis of research on CBT—I, which showed that therapy delivered in digital formats had high acceptability that was comparable to individual and group in-person therapy formats and superior to self-help interventions (Gao et al., 2022). Nonetheless, there remains room for improvement in these ratings of acceptability, especially if it is assumed that those patients who left treatment early might have provided lower ratings on at least some of the questions related to acceptability.

4.5. Strengths and limitations

To our knowledge, this is the first trial to directly test whether including behavioural strategies from CBT-I can increase the effectiveness of a transdiagnostic ICBT course for resolving insomnia symptoms. Importantly, the transdiagnostic ICBT control condition included a sleep resource focused on sleep hygiene and therefore represented a strong control condition. A strength of this trial was the use of an unequal randomization strategy to boost the number of people in the treatment condition available for analysis, which provided greater power to assess how factors affecting sleep and adherence to therapeutic sleep behaviours are related to insomnia symptom change. Including a qualitative review of patient feedback was an important strength of this study, as it allowed for a richer understanding of patient experiences beyond providing an effect size. The patient-oriented approach was valuable for developing the new sleep materials and likely served to increase patient-acceptability, as helpful suggestions about wording and images provided by the patient partners were incorporated into the new sleep resource.

This trial also had several limitations, which should be considered when interpreting the findings. Patients were only assessed for insomnia using a screening measure (the ISI), and a formal diagnosis of insomnia disorder was not made. Although missing data was managed using intention-to-treat analyses, the relatively high rate of missing data at post-treatment and follow-up due to attrition remains another important limitation of the present study because it introduces the possibility of response bias. Although completion rates were similar in both the Sleep-Enhanced and Standard Wellbeing condition, suggesting that whatever factors contribute to a patient leaving the course early were evenly distributed between the two randomly assigned treatment conditions, the chance remains that the observed difference between the Sleep-Enhanced and Standard conditions at post-treatment is caused or inflated by the absence of responses from patients who withdrew because the newly developed intervention did not work for them. Although the weekly self-reports of sleep behaviours were not central to the primary outcomes of this study, it is noted that the weekly nature of these measures limits the accuracy of the data gathered and future studies may wish to consider the possibility that daily sleep diaries or wearable devices may provide more accurate data and serve as a therapeutic reminder to adhere to the behavioural intervention. Another limitation of this study is that the sample size used was appropriate to detect a medium effect size between groups; however, the Wellbeing Course appears to have a reasonably large impact on insomnia symptoms on its own and therefore represents an active comparator. The present trial may have therefore not had suitable power to detect a small effect size between the Sleep-Enhanced and Standard conditions. Another limitation of the study design is that it does not allow us to determine the extent to which therapist support related to insomnia, as opposed to extra treatment materials related to insomnia, resulted in faster improvement in insomnia symptoms, and further research could seek to tease apart the relative effects of these components using a dismantling design. Finally, the results of the current trial may not be generalizable to other transdiagnostic ICBT programs, since ICBT programs often vary in their target population, modality (e.g., web, app, video), and other factors (e.g., duration, amount of therapist support). All data was collected during the COVID-19 pandemic, so additional studies are warranted to examine whether the results are generalizable post-pandemic.

4.6. Future directions

The role of two key behavioural strategies (i.e., sleep restriction, stimulus control) in reducing insomnia symptoms were explored, and the results suggest that sleep restriction may be the more important of the two behavioural strategies when offered within a transdiagnostic ICBT program. However, the observation is correlational only, as the current trial design was not sufficient to establish a causal relationship between engaging in sleep restriction and insomnia relief. A previous study found that multi-component CBT-I leads to greater remission rates than offering sleep restriction or stimulus control independently (Epstein et al., 2012). Future studies could use a dismantling approach to explore the relationship between engaging in sleep restriction and symptom change, as this has not been examined within transdiagnostic ICBT programs. Future studies could also compare different strategies for identifying which patients need to be provided with a sleep intervention and whether larger effects can be obtained if patients receive additional coaching. Furthermore, patient engagement in sleep restriction might be facilitated with a modified version of the technique known as sleep compression, which involves reducing time in bed slowly over a period of weeks (Lichtenstein et al., 2011). Findings from a recent trial comparing sleep restriction and sleep compression suggest that sleep compression may offer similar insomnia symptom reductions (Rosén et al., 2023), and further research could seek to replicate these findings in the context of transdiagnostic ICBT programs. With larger samples, it would be possible to explore what factors predict symptom change and whether some patients with certain characteristics benefit more than others (e.g., Edmonds et al., 2018), and to explore how patient preferences for strategies may impact outcomes (e.g., Hadjistavropoulos et al., 2019).

The exclusion criteria for the current trial were developed so that patients with sleep concerns that were suggestive of a sleep problem other than insomnia (e.g., sleep apnea) would be excluded and directed instead to other more appropriate help for their sleep concerns. Future research investigating the types of education or behavioural interventions that could be included for individuals experiencing sleep apnea and restless leg syndrome could help improve outcomes for these individuals, should they engage with transdiagnostic ICBT. Another potential area of research for improving service to patients with sleep problems would be a collaboration with primary care physicians.

5. Conclusions

Our findings suggest that targeting insomnia as part of a transdiagnostic protocol can offer patients primarily seeking therapy for anxiety and depression additional relief from insomnia symptoms over and above the Wellbeing Course at 8-week follow-up, while still producing large reductions in depression and anxiety. By 20 weeks, however, both the Sleep-Enhanced and Standard transdiagnostic courses had similar large effects for insomnia. The present results suggest that including interventions targeting insomnia in transdiagnostic ICBT may lead to faster improvement in insomnia symptoms and warrants replication in future studies. The rich data produced by the mixed quantitative and qualitative approach of the current trial open avenues for future research. The present trial's results are expected to guide ICBT practice by highlighting the value of targeting symptoms of insomnia as an important part of transdiagnostic treatment models.

Funding

The pilot trial for this study was funded by the Saskatchewan Health Research Foundation and Saskatchewan Centre for Patient-Oriented Research. The Online Therapy Unit is funded by the Saskatchewan Ministry of Health. Author M.E. was funded by Canadian Institutes of Health Research (CIHR) for his doctoral dissertation research. The research was also supported in part by CIHR funding (152917) awarded to H.D.H.. N.T. and B.F.D. are funded by the Australian Government to operate the national MindSpot Clinic. Funders had no involvement in the study design, collection, analysis, or interpretation of the data.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.