The Effects of Digital Cognitive Behavioral Therapy for Insomnia in Chronic Pain: A Randomized Controlled Trial

Abstract

Introduction

Managing chronic pain is a significant healthcare challenge that requires a multimodal treatment approach. In particular, the frequent presence of comorbid insomnia symptoms and its complex bidirectional relationship with pain provides a challenge to adequate treatment. This study was set up to test whether the addition of digital cognitive behavioral therapy for insomnia (dCBT-I) to the regular treatment of chronic pain can improve symptoms in a population comorbid chronic pain and insomnia, when compared to a waitlist control (WLC) condition.

Methods

Participants meeting criteria for chronic pain and insomnia were randomized (1:1) to either dCBT-I or WLC. Assessments took place at baseline, 8 and 24 weeks post-randomization. The primary outcome was insomnia severity (Insomnia Severity Index [ISI]). Secondary outcomes included sleep- and pain-related symptoms as well as measures of well-being and dreams. Linear mixed models were calculated to determine between-group differences.

Results

A total of 207 participants (182 women, mean age = 51.96 ± 12.97 years) were randomized to dCBT-I (n = 103) or WLC (n = 104). The dCBT-I group showed large improvements in the severity of insomnia compared to the WLC both after 8 (−4.36, p < 0.001; d = −1.18) and 24 weeks (−4.88, p < 0.001; d = −1.32). Pain-related impairments and life control also improved, favoring dCBT-I (ds = 0.22–0.35). Moderate-to-large treatment effects were also observed for secondary outcomes, including daytime sleepiness, fatigue, and dysfunctional beliefs and attitudes about sleep (ds = 0.47–1.12).

Conclusion

This study confirmed the effects of dCBT-I in reducing the severity of insomnia in individuals with chronic pain and insomnia. Improvements in symptoms of chronic pain further emphasize the potential of dCBT-I as a scalable, evidence-based intervention to address the complex challenges of dual pathology in clinical practice, and it presents a promising extension of multimodal pain management.

Introduction

Chronic pain is a widespread and challenging disease, affecting up to 30% of the population worldwide [1]. According to the Global Burden of Disease Study [2], pain and pain-related illnesses are the main cause for disability and burden of disease worldwide, and one of the most common reasons for seeking medical care [3]. In Germany, chronic pain disorders cause annual costs of around EUR 38 billion, which include treatment costs, sick pay, loss of working hours, and early retirement [4]. According to the latest definition by the International Association for the Study of Pain (IASP), chronic pain is defined as pain that persists or recurs for longer than 3 months [5]. It is characterized by the loss of its original protective and warning function, evolving into an independent condition that persists even in the absence of an acute underlying cause [6]. Today, the term “chronic pain” is used as an umbrella term to describe a wide range of painful conditions [7]. This includes pain with or without an identifiable medical cause, as well as mixed forms, reflecting the heterogeneity and complexity of this clinical condition.

Insomnia, one of the most prevalent and debilitating sleep disorders, used to be seen as a direct consequence of chronic pain [8]. Today, however, it is widely recognized that insomnia and chronic pain can co-occur and exacerbate each other [9]. The prevalence of insomnia in patients with chronic pain ranges from 50.4% to 72.0%, with insomnia being the most common comorbid sleep disorder for chronic pain [10, 11]. A vicious cycle is proposed, whereby poor sleep leads to more pain during the day and more pain results in poorer subsequent sleep [12, 13].

Regarding the bidirectional link between insomnia and chronic pain, a detailed review showed that insomnia is a stronger predictor of chronic pain onset than the other way around [14]. This finding was also confirmed by longitudinal data, which showed that sleep symptoms have a greater influence on pain [15].

Experimental and clinical evidence suggests that restorative sleep enhances descending pain inhibition, lowers levels of pro-inflammatory cytokines that contribute to central sensitization, and stabilizes neurotransmitter systems involved in pain processing [16–18]. From a psychological perspective, several theoretical models offer insights into the co-occurrence and mutual maintenance of insomnia and chronic pain. The fear-avoidance model of chronic pain [19, 20] posits that catastrophic interpretations of pain lead to fear, hypervigilance, avoidance behavior, and ultimately disability. Neurocognitive models, such as the triple network model [21], suggest that chronic pain is associated with dysfunctional connectivity between three large-scale brain networks, the salience network, default mode network, and central executive network. These alterations impair emotional regulation, threat detection, and cognitive control.

Building on these frameworks, a recently proposed cognitive model of comorbid insomnia and chronic pain emphasizes the role of generalized cognitive biases, such as attentional hypervigilance, negative expectations, and maladaptive belief systems, that interact across pain and sleep domains [22]. The model posits that such transdiagnostic biases contribute to the progression from acute to chronic symptomatology by amplifying physiological arousal and reinforcing avoidance-based coping strategies. Notably, these biases are not restricted to one domain (e.g., pain catastrophizing or sleep-related worry), but rather generalize across somatic experiences, leading to increased perceived threat, functional impairment, and emotional distress. Consequently, improving sleep quality and duration may attenuate both physiological and psychological mechanisms that sustain chronic pain, thereby reducing symptom burden and improving overall functioning. Targeting shared cognitive-affective mechanisms, such as fear, hypervigilance, and avoidance, may offer a promising approach for addressing comorbid insomnia and chronic pain. Addressing insomnia in the management of chronic pain is therefore not only justified, but essential.

A multimodal therapy approach, as part of a biopsychosocial model, is recommended for the treatment of chronic pain [23, 24]. Additionally, the Centers for Disease Control and Prevention (CDC) guidelines strongly encourage non-pharmacological pain management as the first-line treatment for chronic pain [25]. However, in Germany, around 50% of affected individuals receive monotherapy (i.e., mostly non-steroidal anti-inflammatory drugs or opioids) [26] with no emphasis on addressing insomnia in their treatment [27].

Effective treatment of insomnia is typically achieved through cognitive behavioral therapy for insomnia (CBT-I), which integrates multiple therapeutic components in a multimodal approach. While individual core components, such as sleep restriction or stimulus control, have demonstrated promising effects as standalone interventions, the combined multimodal approach has generally proven to be superior [28]. Owing to its robust treatment effects across various conditions (Hedges’ g = 0.98 [29]), CBT-I is recommended as the first-line treatment for insomnia [30].

Its effectiveness extends to comorbid chronic pain, where CBT-I has shown superiority over cognitive behavioral therapy for chronic pain and hybrid approaches [31, 32]. Furthermore, consistent findings highlight improvements in both short- and long-term sleep quality, as well as relief from sleep-related symptoms, across various cancer-related and chronic nonmalignant pain conditions [33–38].

Despite strong evidence supporting CBT-I, only a minority of affected patients receive adequate psychotherapeutic treatment for insomnia [39]. Since psychotherapy is often initiated long after diagnosis, this delay may worsen insomnia and hinder recovery [40]. In this context, digital implementations of CBT (dCBT-I) offer a promising solution, providing timely, guideline-based treatment to a larger number of patients in need [41]. Research on the effectiveness of digital CBT-I has shown similar positive effects on insomnia symptoms compared to face-to-face interventions [42–44].

To date, research on dCBT-I in individuals with comorbid chronic pain is still limited. Two studies have examined this population, including a single-arm study using digitally delivered therapist sessions [45] and a small-scale randomized controlled trial of an automated program [46]. Both studies reported improvements in insomnia severity, although neither demonstrated statistically significant effects on pain-related outcomes. These findings underscore the need for further investigation of fully automated dCBT-I interventions in this patient group. The present study addresses this gap by evaluating the effectiveness of a fully self-guided dCBT-I program without therapist contact in individuals with comorbid insomnia and chronic pain.

This study aimed to investigate whether the addition of an effective dCBT-I (somnio, mementor DE GmbH, Leipzig, Germany) to the regular treatment of chronic pain can improve symptoms in a population comorbid chronic pain and insomnia. The effectiveness of this digital intervention has been demonstrated in previous studies, showing large effect sizes (ESs) comparable to those of face-to-face CBT-I [47–50], supporting its potential as a viable treatment component in this context. The primary study hypothesis is that dCBT-I leads to a reduction in self-reported insomnia severity compared to the WLC at 8 (primary time point) and 24 weeks (follow-up). To investigate whether the treatment of insomnia has spill-over effects on related conditions (pain symptoms in particular), potential underlying mechanisms, and overall well-being, it was further hypothesized as follows:

• (1)dCBT-I leads to a reduction in the severity of sleep-related outcomes compared to the WLC
• (2)dCBT-I leads to an improvement in pain-related outcomes compared to the WLC
• (3)dCBT-I leads to an improvement in well-being and quality of life compared to the WLC
• (4)dCBT-I leads to an improvement in anxiety and depression compared to the WLC
• (5)dCBT-I leads to a reduction in nightmare distress compared to the WLC.

Additionally, adherence, treatment satisfaction, and adverse events were obtained exploratively to evaluate treatment experience.

Methods

Study Design

We conducted a two-armed, randomized controlled trial, assigning adult participants randomly to either the intervention or the waitlist control (WLC) group. While the intervention group received 8 weeks of dCBT-I (somnio, mementor DE GmbH, Leipzig, Germany), the WLC group received no additional treatment during this period but could access dCBT-I after study completion. Both groups were not limited in their access to regular care (care-as-usual [CAU]). A CAU (+WLC) control group was chosen to allow the evaluation of the added benefit of the dCBT-I intervention over and above the standard care currently available. It reflects the real-world clinical practices that participants would otherwise receive if they were not enrolled in the trial and comparing to CAU allows determining whether the novel approach offers a meaningful improvement over existing care. Assessments took place at baseline, 8 weeks (posttreatment), and 24 weeks (follow-up) post-randomization. The reporting of this study followed the CONSORT guideline [51, 52] (see CONSORT checklist in online suppl. material at https://doi.org/10.1159/000547436).

Participants and Procedure

The recruitment phase took place from January to December 2023, with participants primarily reached through advertisements on social media platforms such as Facebook and Instagram. To be eligible, participants had to meet the following criteria: (1) be at least 18 years old, (2) have a diagnosis of chronic insomnia according to DSM-5 guidelines [53], confirmed through a diagnostic interview and an Insomnia Severity Index (ISI) score of ≥10 [54], (3) have a concurrent diagnosis of chronic pain disorder with both somatic and psychological factors according to ICD-10 [55], verified by a diagnostic interview, (4) be proficient in using digital devices such as smartphones, tablets, or computers, (5) have access to a stable internet connection, and (6) sufficient German language skills.

Exclusion criteria included (1) a diagnosis of bipolar disorder or psychosis, (2) habitual alcohol consumption (≥3 glasses daily for at least 3 weeks), cannabis use (≥1 time per week), or use of other illicit substances, (3) any indication of suicidality, or (4) a history of epilepsy. Once potential participants expressed interest in the study, they were directed to an online screening via the SoSci Survey platform (SoSci Survey GmbH, Munich, Germany). First, they were instructed to carefully review the provided patient information and give their consent to participate in the screening process. The online screening then assessed whether they met the study’s inclusion and exclusion criteria. Participants who appeared eligible were asked to submit their contact information so the research team could schedule a telephone interview. During the interview, the study was explained in detail once again, providing an opportunity to address any questions and outline the next steps. Additionally, a clinical interview was conducted to reassess the inclusion and exclusion criteria, confirming participant eligibility. This step also verified the diagnoses of chronic insomnia (according to DSM-5) and chronic pain disorder (according to ICD-10). Other psychological or sleep-related disorders were not systematically assessed during the interview. The interviews were conducted by advanced psychology master’s students under the supervision of a licensed psychotherapist. Following successful completion of the interview, participants who met all criteria received a link to the baseline assessment.

At the start of the baseline assessment, participants were required to provide informed consent before proceeding to the baseline questionnaires. Once the assessment was completed, they were randomly assigned to either the intervention or control group. Participants in the intervention group received access to the 8-week dCBT-I program and were instructed to begin immediately. Those in the WLC group received no additional treatment. Both groups were encouraged to continue their regular medical care throughout the study period.

The posttreatment assessment took place online 8 weeks after randomization, followed by the follow-up assessment 24 weeks after randomization, also conducted online. At this point, study participation was complete, and the WLC group was subsequently granted access to the dCBT-I program. There was no financial compensation for participation.

The dCBT-I Intervention

Participants in the intervention group were given access to the dCBT-I program somnio (mementor DE GmbH, Leipzig, Germany). somnio consists of 10 core modules based on guideline-recommended CBT-I techniques, which are sequentially unlocked as participants complete diary entries and previous modules. The modules cover essential CBT-I components, including psychoeducation, relaxation techniques, stimulus control, sleep restriction therapy, and cognitive therapy. Follow-up modules are designed to reinforce learned content and reduce the risk of relapse. The program is delivered through an automated, interactive avatar that guides participants through the intervention. For a comprehensive overview of somnio’s structure, see Table 1. The program’s effectiveness has been validated in three randomized controlled trials [47–49], as well as in a retrospective user data analysis of regular care patients [56] and a subgroup analysis of participants with high baseline depressive and anxiety scores [57].

Table 1.

Overview of module content in the dCBT-I intervention somnio

Module	Description
01 Introduction	The digital sleep expert “Albert” conducts an initial assessment of users’ sleep patterns and problems, and introduces the structure and goals of the intervention
02 Sleep journal	Users are introduced to the sleep diary. It includes morning and evening logs which enable analysis of sleep patterns, such as bedtime, total sleep time, and sleep efficiency
03 Sleep knowledge	This module offers evidence-based information on sleep, while addressing misconceptions and reducing sleep-related anxiety and dysfunctional beliefs
04 Practical exercise	Previously learned content is consolidated and applied using practical examples
05 Cycle of insomnia	Users gain insight into the definition, prevalence, and causes of sleep disorders and develop a personalized model of their own “cycle of insomnia”
06 Sleeping hours	This module introduces sleep restriction as a core treatment strategy. Using data from the sleep diary, a personalized sleep window is recommended
07 Relaxation	Users are introduced to Jacobson’s PMR and guided in its practical application
08 Sleep behavior	This module examines psychological and behavioral influences on sleep, addresses common misconceptions, and introduces stimulus control to rebuild a healthy connection between bed and sleep
09 Thoughts	Users engage in cognitive restructuring by identifying and challenging dysfunctional sleep-related beliefs. Through case examples, they learn to recognize unhelpful thoughts, correct misconceptions, and develop more realistic, sleep-promoting thinking patterns
10 Everyday decisions	This module helps users identify everyday situations that trigger safety behaviors and teaches more effective coping strategies to support better sleep
11 Closing session	In the final module, users review key concepts from the program and reinforce their knowledge through a summary and interactive quiz
12 Follow-up	Key sleep parameters and insomnia symptoms are monitored regularly to ensure continuous assessment throughout the program

PMR, progressive muscle relaxation.

Randomization and Masking

Participants were assigned to either dCBT-I or WLC in a 1:1 ratio using a randomization sequence generated online (Sealed Envelope, London, UK), utilizing varying block sizes (2–4) to ensure concealment of future allocations. The randomization sequence was only accessible to a designated member of the study team who had no interaction with the participants throughout the study period. Participants were provided with general information about the different groups and were informed of their group allocation. However, they were not made aware of the study’s specific hypotheses. The study team was informed of participant group assignments after randomization but had no influence on the intervention and the digital data collection. All data were blinded before the analysis.

Measurements

• Primary Outcome

  • Insomnia Severity. Self-reported insomnia severity was assessed using the ISI [54], a 7-item measure rated on a 5-point Likert scale (0 = not at all to 4 = extremely), reflecting the past 2 weeks. Total scores range from 0 to 28, with higher scores indicating greater severity. The ISI has demonstrated strong internal consistency [58–60]. Regarding clinical significance, responder rates (defined as a reduction in ISI ≥8) and remission rates (ISI total score <8) were calculated [54, 58].

• Secondary Outcomes

  • Sleep-Related Outcomes. Daytime Sleepiness: Daytime sleepiness was evaluated using the Epworth Sleepiness Scale (ESS) [61, 62], an 8-item questionnaire rated on a 4-point Likert scale (0 = no chance of dozing to 3 = high chance of dozing). The total score ranges from 0 to 24, with higher scores indicating greater sleepiness. The ESS is a well-validated instrument, with strong internal consistency (α = 0.73–0.86) [63].

  • Fatigue. Fatigue was assessed using the Fatigue Severity Scale (FSS), a 9-item instrument measuring fatigue over the past week. Items are rated on a 7-point Likert scale (1 = strong disagreement to 7 = strong agreement), with higher mean scores indicating greater fatigue. The FSS is a validated measure of fatigue severity [64].

  • Dysfunctional Beliefs and Attitudes about Sleep. The short version of the Dysfunctional Beliefs and Attitudes About Sleep Scale (DBAS-16) identifies maladaptive sleep-related cognitions [65, 66]. The 16 items are rated on an 11-point Likert scale (0 = strongly disagree to 10 = strongly agree). The DBAS-16 is a valid and reliable instrument [67].

Pain-Related Symptoms

Pain-Related Symptoms. The German Pain Questionnaire (DSF) [68] captures multidimensional aspects of chronic pain. Pain intensity and pain-related impairment were assessed using the Chronic Pain Grade Questionnaire (CPG) [69], a 7-item self-report measure rated on an 11-point Likert scale (0 = no pain/impairment to 10 = worst imaginable pain/complete impairment). Pain intensity and impairment scores are each calculated by multiplying the mean of three items by 10. Scores ≥50 indicate high pain intensity or impairment. The CPG shows strong internal consistency (α = 0.91) and is considered reliable and valid [70]. Severity grades (0–4) were calculated based on pain intensity, impairment, and number of disability days, following von Korff’s method [69].

Impact of Chronic Pain (Multidimensional Pain Inventory [MPI]). The MPI was used to measure the impact of chronic pain [71]. It comprises three sections with distinct subscales. The first section deals with the impact of pain on a person’s life and contains 5 subscales: pain severity, interference, life control, affective distress, and support. Items are rated on a 7-point Likert scale with varying anchors. Higher scores on pain severity, interference, and affective distress indicate greater impairment; higher scores on support and life control indicate less impairment. The MPI is a reliable and valid tool for chronic pain assessment [72].

Well-Being and Quality of Life

Well-Being. Well-being was measured using the World Health Organization-Five Well-being Index (WHO-5) [73, 74], a 5-item measure rated on a 6-point Likert scale (0 = none of the time to 5 = all the time). Higher scores indicate greater well-being. The WHO-5 has demonstrated excellent psychometric properties [75].

Quality of Life. Quality of life was assessed using the brief version of the World Health Organization Quality of Life Questionnaire (WHOQOL-BREF) [76], a 26-item self-report questionnaire rated on a 5-point Likert scale. It covers four domains: physical health, psychological health, social relationships, and environment. Domain scores range from 4 to 20, with higher scores indicating better quality of life. The WHOQOL-BREF is a reliable and valid instrument, including in German-speaking populations [74, 76].

Frustration and Satisfaction of Psychological Needs. The Basic Psychological Need Satisfaction and Frustration Scale (BPNSFS) [77] assesses satisfaction and frustration of the three basic psychological needs: autonomy, competence, and relatedness. It includes 24 items rated on a 5-point Likert scale (1 = completely disagree to 5 = completely agree), with 12 items each for satisfaction and frustration. The BPNSFS is a reliable and valid instrument in diverse populations [78].

Motivational Incongruence of Approach and Avoidance Goals. The Incongruence Questionnaire (K-INK) [79], based on Grawe’s basic needs theory [80], assesses motivational incongruence via two subscales: approach (14 items) and avoidance (9 items). Items are rated on a 4-point Likert scale (1 = far too little to 4 = fully sufficient). The K-INK is a reliable and valid instrument [79].

Depressive and Anxiety Symptoms

Depressive Symptoms. Depressive symptoms were assessed using the Allgemeine Depressionsskala Kurzversion (ADS-K) [81], the short form of the German version of the Center for Epidemiological Studies Depression Scale (CES-D) [82]. The 15 items assess symptoms over the past week on a 4-point Likert scale (1 = rarely to 4 = most of the time). Total scores range from 0 to 45, with higher scores indicating greater severity. The ADS-K has shown strong internal consistency (α = 0.88–0.95) [83].

Anxiety Symptoms. The trait version of the State-Trait Anxiety Inventory (STAI-T) was used to assess anxiety symptomatology [84, 85], comprising 20 items rated on a 4-point Likert scale (1 = almost never to 4 = almost always). Total scores range from 20 to 80, with higher scores indicating greater anxiety. The STAI-T shows good to excellent internal consistency (α = 0.89–0.91) [86].

Dreams and Nightmares

Dream Recall Frequency, Nightmare Frequency, and Nightmare Distress. Dream recall frequency and nightmare frequency were assessed with an open-ended question: “how many dreams/nightmares have you had in the past 4 weeks?” Nightmare distress was measured using the Nightmare Distress Questionnaire (NDQ) [87], a 13-item scale assessing general distress, sleep impact, and daytime effects. Most items are rated on a 5-point Likert scale (1 = never to 5 = always). The NDQ has shown good reliability (α = 0.80) [88].

Adherence, Treatment Satisfaction, and Adverse Events

At posttreatment, participants in the dCBT-I group reported program usage via a multiple-choice item (0–10 modules completed). A module was counted as completed if it had been fully worked through. Treatment satisfaction was assessed using three 5-point Likert scales evaluating overall satisfaction, expectation fulfillment, and adherence (1 = not at all to 5 = completely). Participants were instructed to report adverse events to the study team.

Data Analysis

The sample size was calculated using G*Power [89], with α = 0.05 and power of 90%, to an ES of d = 0.48 [46] on the primary outcome at posttreatment (week 8). Power analysis revealed a sample size of N = 186 participants, which would be needed to detect between-group effects. Accounting for 10% attrition [48], we sought to recruit 204 participants (102 in each group). All statistical analyses were conducted using SPSS version 29 (IBM). Consistent with CONSORT guidelines, the analysis followed the intention-to-treat principle, including available data from all randomized participants [51, 52, 90].

Linear mixed-effects regression models were applied to assess between-group comparisons in primary and secondary outcomes, incorporating fixed effects for group and time. These models accounted for missing data, with outcomes at 8 and 24 weeks entered as response variables. Baseline measurements were included as covariates to control for any initial differences, and a participant-specific random intercept was added to account for repeated measures [91]. A group-by-time interaction term was included to estimate between-group effects at each time point. The covariance structure was set to unstructured.

Cohen’s d was employed to quantify between-group ESs, calculated by dividing the adjusted mean difference by the standard deviation of both groups at baseline [92]. Between-group differences in these dichotomous outcomes were assessed using Pearson’s chi-squared tests, and ESs for categorical outcomes were quantified using the phi coefficient. Descriptive statistics are reported as unadjusted means (M) and standard deviations, while frequencies are presented for binary outcomes.

Results

A total of N = 584 potential participants completed the online screening process. Of these, n = 111 did not complete the full screening, and n = 174 met at least one exclusion criterion, resulting in n = 299 participants being invited for a subsequent telephone interview. During this phase, n = 83 participants were excluded, primarily due to unsuccessful contact attempts, leaving n = 216 participants, who were deemed eligible. Of these, N = 207 participants completed the baseline assessment and were subsequently randomized. The dropout rate at the posttreatment assessment was 8.70%, increasing slightly to 11.59% at follow-up.

The participant flow is shown in Figure 1. Participants reported an average age of 51.96 ± 12.97 years. The sample predominantly consisted of women (n = 182, 87.9%), with the majority having either completed an apprenticeship (33.3%) or holding a university degree (29.5%). On average, participants reported experiencing insomnia for 9.78 ± 9.83 years and chronic pain for 12.86 ± 11.32 years. The most commonly reported pain condition was back pain (n = 136, 65.7%), followed by joint pain (n = 129, 62.3%) and muscle pain (n = 126, 60.9%). Abdominal pain was the least frequently reported condition, affecting 26.6% (n = 55) of participants. Notably, 21.4% (n = 22) of participants in the intervention group and 18.3% (n = 19) in the WLC group reported currently receiving psychotherapeutic treatment. Participants had an average ISI score of 18.70 ± 3.69, which falls within the range of moderate insomnia severity. Detailed demographic characteristics of both groups are provided in Table 2.

Fig. 1.

Participant flow.

Table 2.

Participant demographic characteristics

​	dCBT-I	Control group
	n = 103	n = 104
Baseline characteristics
Age, M (SD), years	49.78 (13.59)	54.13 (12.02)
Female, n (%)	89 (86.40)	93 (89.40)
Symptom duration insomnia, M (SD), years	9.23 (9.29)	10.32 (10.35)
Symptom duration pain, M (SD), years	11.36 (10.85)	14.32 (11.62)
Distinct trigger that caused insomnia symptoms, n (%)	22 (21.40)	31 (29.80)
Shared bedroom, n (%)	36 (35.00)	50 (48.10)
Treatments
Psychotherapy
Current, n (%)	22 (21.40)	19 (18.30)
Former, n (%)	47 (45.60)	59 (56.70)
Medicationa
CNS medication, n (%)	32 (31.10)	41 (39.40)
Sleep medication, n (%)	17 (16.50)	26 (25.00)
Other medication, n (%)	71 (68.90)	70 (67.30)
Further treatmentsa
General practitioner	74 (71.80)	80 (76.90)
Surgeon	2 (1.90)	6 (5.80)
Neurosurgeon	4 (3.90)	8 (7.70)
Alternative practitioner	6 (5.80)	9 (8.70)
Internist	13 (12.60)	18 (17.30)
Neurologist	38 (36.90)	39 (37.50)
Orthopedist	54 (52.40)	41 (39.40)
Psychiatrist	17 (16.50)	18 (17.30)
Pain therapist	29 (28.20)	32 (30.80)
Physiotherapist	34 (33.00)	31 (29.80)
Osteopath	13 (12.60)	15 (14.40)
Comorbidities
Physical illnesses, n (%)	99 (96.10)	103 (99.00)
Type of chronic paina
Fibromyalgia	26 (25.20)	42 (40.40)
Headache	42 (40.80)	45 (43.30)
Back pain	64 (62.10)	72 (69.20)
Muscle pain	58 (56.30)	68 (65.40)
Joint pain	66 (64.10)	63 (60.60)
Stomach pain	22 (21.40)	33 (31.70)
Nerve pain	38 (36.90)	41 (39.40)
Sleep disorders, n (%)
Sleepwalking	3 (2.90)	0 (0.00)
Narcolepsy	0 (0.00)	0 (0.00)
Pavor nocturnus	1 (1.00)	0 (0.00)
Bruxism	32 (31.10)	34 (32.70)
Nightmare disorder	3 (2.90)	3 (2.90)
Restless legs	14 (13.60)	14 (13.50)
Obstructive sleep apnea	18 (17.50)	21 (20.20)
Psychological diagnoses
Current, n (%)	19 (18.40)	19 (18.30)
Thereof major depression, n (%)	14 (73.68)	12 (63.16)
Former n (%)	47 (45.60)	46 (44.20)
Thereof major depression, n (%)	36 (76.60)	38 (82.61)

M, means; SD, standard deviation; n, the number of participants.

^aMultiple selection was possible.

Primary Outcome

Insomnia Severity

The linear mixed-model analysis revealed large between-group ESs favoring the intervention group at both 8 weeks (p < 0.001, d = −1.18) and 24 weeks (p < 0.001, d = −1.32). On average, insomnia symptom severity decreased by 4.36 points at 8 weeks and 4.88 points at 24 weeks in the intervention group compared to the WLC group. A graphical representation of these results can be found in Figure 2.

Fig. 2.

Changes in primary outcome (insomnia severity, measured with the ISI), across both groups and all assessments. Unadjusted means (±1 SD) are presented for both groups. Statistical group differences are derived from linear mixed models and represented by a double asterisk (**p < 0.001).

After 8 weeks, 31.07% (n = 32) of participants in the intervention group were classified as responders, compared to 3.85% (n = 4) in the WLC (χ² [1, n = 189] = 29.57, p < 0.001, φ = −0.40). By 24 weeks post-randomization, 35.92% (n = 37) of the intervention group were classified as responders, compared to 7.69% (n = 8) in the WLC (χ² [1, n = 185] = 25.96, p < 0.001, φ = −0.38).

Regarding remission rates, 19.42% (n = 20) of participants in the intervention group and 1.92% (n = 2) of participants in the WLC met the remission criteria after 8 weeks (χ² [1, n = 189] = 18.24, p < 0.001, φ = −0.31). At 24 weeks, 15.53% (n = 16) of participants in the intervention group met the remission criteria, compared to 2.88% (n = 3) in the WLC (χ² [1, n = 185] = 10.39, p = 0.001, φ = −0.24). Detailed baseline statistics and statistical results for the primary and sleep-related secondary outcomes are presented in Table 3.

Table 3.

Between-group effects of dCBT-I versus control group on primary and other sleep-related secondary outcomes

​	dCBT-I		Control group		Diffadj	p value	95% CI		ES
	M	SD	M	SD
Insomnia severity (ISI)
Baseline	18.47	3.72	18.92	3.67	​	​	​	​	​
Week 8	12.64	5.70	17.30	4.09	−4.36	<0.001	−5.36	−3.37	−1.18
Week 24	12.07	5.30	17.16	4.09	−4.88	<0.001	−5.88	−3.88	−1.32
Daytime sleepiness (ESS)
Baseline	9.45	5.00	8.97	4.44	​	​	​	​	​
Week 8	8.11	4.60	9.84	4.63	−2.24	<0.001	−2.93	−1.56	−0.47
Week 24	7.89	4.20	10.11	4.74	−2.82	<0.001	−3.51	−2.13	−0.60
Fatigue (FSS)
Baseline	5.39	1.01	5.51	1.08	​	​	​	​	​
Week 8	4.59	1.42	5.47	1.24	−0.78	<0.001	−1.03	−0.54	−0.75
Week 24	4.60	1.47	5.49	1.17	−0.82	<0.001	−1.07	−0.57	−0.78
Dysfunctional beliefs and attitudes about sleep (DBAS-16)
Baseline	6.14	1.33	5.97	1.57	​	​	​	​	​
Week 8	5.00	1.62	6.28	1.51	−1.49	<0.001	−1.77	−1.21	−1.03
Week 24	4.78	1.66	6.16	1.50	−1.63	<0.001	−1.92	−1.35	−1.12

ISI, Insomnia Severity Index (range: 0–28); ESS, Epworth Sleepiness Scale (range: 0.24); FSS, Fatigue Severity Scale (range: 1–7); DBAS-16, Dysfunctional Beliefs and Attitudes about Sleep Scale (range: 0–10); Diff_adj, adjusted mean difference derived from linear mixed model; 95% CI, 95% confidence interval of the adjusted mean difference; ES, effect size (Cohen’s d); M, unadjusted means; SD, standard deviation.

Significant p values are displayed in bold.

Sleep-Related Secondary Outcomes

Daytime Sleepiness

The analysis of daytime sleepiness revealed small to medium ESs between both groups at 8 (p < 0.001, d = −0.47) and 24 weeks (p < 0.001, d = −0.60) post-randomization in favor of the dCBT-I group.

Fatigue

Between-group comparisons demonstrated medium-sized treatment effects in favor of the dCBT-I group at both 8 weeks (p < 0.001, d = −0.75) and 24 weeks (p < 0.001, d = −0.78) post-randomization, indicating a significant reduction in fatigue attributable to the dCBT-I intervention.

Dysfunctional Beliefs and Attitudes about Sleep

The linear mixed-model analysis identified large ESs favoring the digital CBT-I group in reducing dysfunctional beliefs and attitudes about sleep at both 8 weeks (p < 0.001, d = −1.03) and 24 weeks (p < 0.001, d = −1.12).

Pain-Related Secondary Outcomes

Pain-Related Symptoms

The results of the linear mixed model showed small ESs in relation to the pain intensity of the CPG at 8 (p < 0.001, d = −0.37) and 24 weeks (p = 0.035, d = −0.23) post-randomization. The dCBT-I group showed a reduced pain intensity compared to the WLC. A similar pattern of results was shown for the experience of impairment. The dCBT-I group reported reduced impairment with small ESs after 8 (p = 0.004, d = −0.29) and 24 weeks (p < 0.001, d = −0.35) compared to the WLC. The dCBT-I group also reported a reduction in the degree of chronification compared to the WLC with small ESs at both time points (p = 0.017, d = −0.24; and p = 0.002, d = −0.32). See Figure 3a–c for graphical representation of these results.

Fig. 3.

Changes in secondary pain-related outcomes (a pain intensity, b experience of impairment, and c degree of chronification) across both groups and all assessments. Unadjusted means (±1 SD) are presented for both groups. Statistical group differences are derived from linear mixed models and represented by a (double) asterisk (*p < 0.05, **p < 0.001).

Impact of Chronic Pain

The MPI showed small treatment effects in favor of the dCBT-I group both after 8 weeks (p < 0.001, d = −0.33) and after 24 weeks (p = 0.001, d = −0.31) with regard to the subscale Interference. Significant improvements with small ESs were also found for the subscale Life Control in the dCBT-I group compared to the WLC at both time points (p = 0.044, d = 0.22; and p = 0.030, d = 0.24). In contrast, no significant treatment effects between both groups were found on the subscales Pain Severity, Affective Distress, and Support at any time point (all ps > 0.005). See Table 4 for statistical analyses of all pain-related secondary outcomes.

Table 4.

Between-group effects of dCBT-I versus control group on pain-related secondary outcomes

​	dCBT-I		Control group		Diffadj	p value	95% CI		ES
	M	SD	M	SD
Pain-related symptoms
Pain intensity (DSF)
Baseline	62.33	14.29	63.01	15.86	​	​	​	​	​
Week 8	58.20	17.67	64.56	16.55	−5.55	<0.001	−8.69	−2.41	−0.37
Week 24	59.33	17.43	63.80	17.17	−3.42	0.035	−6.60	−0.24	−0.23
Experience of impairment (DSF)
Baseline	61.97	21.10	63.11	22.47	​	​	​	​	​
Week 8	55.51	25.43	62.55	25.09	−6.25	0.004	−10.47	−2.03	−0.29
Week 24	54.85	25.02	63.12	24.52	−7.56	<0.001	−11.82	−3.29	−0.35
Degree of chronification (v. Korff)
Baseline	3.22	1.03	3.19	1.03	​	​	​	​	​
Week 8	3.01	1.15	3.21	1.03	−0.25	0.017	−0.45	−0.05	−0.24
Week 24	2.96	1.18	3.22	0.98	−0.33	0.002	−0.53	−0.12	−0.32
Impact of chronic pain
Pain severity (MPI)
Baseline	11.20	2.81	12.26	3.02	​	​	​	​	​
Week 8	10.89	3.40	12.05	3.24	−0.41	0.230	−1.08	0.26	−0.14
Week 24	10.96	3.53	11.86	3.20	−0.14	0.682	−0.82	0.53	−0.05
Interference (MPI)
Baseline	39.18	10.17	40.44	11.95	​	​	​	​	​
Week 8	35.27	13.19	39.89	13.47	−3.68	<0.001	−5.74	−1.61	−0.33
Week 24	35.54	13.61	39.87	13.55	−3.45	0.001	−5.53	−1.36	−0.31
Affective distress (MPI)
Baseline	9.09	2.37	9.27	2.12	​	​	​	​	​
Week 8	8.93	2.34	9.41	2.28	−0.29	0.274	−0.81	0.23	−0.13
Week 24	8.92	2.74	9.32	2.56	−0.29	0.276	−0.82	0.23	−0.13
Support (MPI)
Baseline	9.52	4.96	9.97	4.50	​	​	​	​	​
Week 8	8.42	4.91	8.63	5.19	0.17	0.656	−0.58	0.92	0.04
Week 24	8.73	4.98	8.87	5.21	0.01	0.979	−0.75	0.77	0.00
Life control (MPI)
Baseline	10.06	3.51	9.97	3.34	​	​	​	​	​
Week 8	10.29	3.75	9.31	3.39	0.75	0.044	0.02	1.48	0.22
Week 24	9.97	4.26	8.78	3.71	0.82	0.030	0.08	1.55	0.24

DSF, Deutscher Schmerzfragebogen (German Pain Questionnaire); pain intensity range: 0–100; experience of impairment range: 0–100; degree of chronification range: 0–4; MPI, Multidimensional Pain Inventory; pain severity range: 0–18; interference range: 0–60; affective distress range: 0–18; support range: 0–18; life control range: 0–18; Diff_adj, adjusted mean difference derived from linear mixed model; 95% CI, 95% confidence interval of the adjusted mean difference; ES, effect size (Cohen’s d); M, unadjusted means; SD, standard deviation.

Significant p values are displayed in bold.

Secondary Outcomes Relating to Well-Being and Quality of Life

Well-Being

Small treatment effects were found regarding well-being between the dCBT-I group and the WLC after 8 (p = 0.007, d = 0.30) and 24 weeks (p = 0.004, d = 0.34) post-randomization. Compared to the WLC, the dCBT-I group showed improved well-being.

Quality of Life

Quality of life was analyzed by domain and showed small treatment effects for physical health in favor of the dCBT-I group at both 8 (p < 0.001, d = 0.45) and 24 weeks (p < 0.001, d = 0.39) post-randomization. With regard to psychological health, no treatment effect was observed between the groups 8 weeks post-randomization (p = 0.172). After 24 weeks, however, there was a small treatment effect in favor of the dCBT-I group (p = 0.029, d = 0.17). No effects between both groups were found at any time point in the domains social relationships and environment (ps = 0.253–0.855).

Satisfaction and Frustration of Psychological Needs

Satisfaction and frustration of psychological needs were analyzed for each subscale. With regard to the satisfaction subscales, no significant between-group effects were found at any time point (all ps = 0.142–0.902). Regarding autonomy frustration, the dCBT-I group showed reduced frustration after 8 weeks compared to the WLC group with a small ES (p = 0.011, d = −0.23). This effect did not remain stable after 24 weeks (p = 0.053). Concerning relatedness frustration, the WLC group showed reduced frustration after 8 weeks compared to the dCBT-I group with a small ES (p = 0.019, d = 0.21). This effect did not remain stable after 24 weeks (p = 0.766). In terms of competence frustration, the dCBT-I group showed reduced frustration with small ESs compared to the WLC group after 8 (p = 0.002, d = −0.25) and 24 weeks (p = 0.014, d = −0.20) post-randomization.

Motivational Incongruence

No treatment effects were found between the dCBT-I group and the WLC in terms of motivational incongruence of approach goals at 8 (p = 0.124) and 24 weeks (p = 0.168) post-randomization. A small between-group effect in favor of the dCBT-I group was found for motivational incongruence of avoidance goals 8 weeks after randomization (p = 0.027, d = −0.24). After 24 weeks, however, no significant difference was found between both groups (p = 0.864). See Table 5 for statistical analyses of all well-being and quality of life-related secondary outcomes.

Table 5.

Between-group effects of dCBT-I versus control group relating to well-being and quality of life

​	dCBT-I		Control group		Diffadj	p value	95% CI		ES
	M	SD	M	SD
Well-being (WHO-5)
Baseline	7.70	4.19	7.37	4.42	​	​	​	​	​
Week 8	8.60	5.77	7.04	4.86	1.31	0.007	0.35	2.27	0.30
Week 24	8.72	5.88	7.07	4.74	1.45	0.004	0.47	2.42	0.34
Quality of life – physical health (WHOQOL-BREF)
Baseline	10.31	2.52	10.05	2.91	​	​	​	​	​
Week 8	11.32	2.82	9.83	2.93	1.23	<0.001	0.78	1.67	0.45
Week 24	11.30	3.19	9.82	3.09	1.07	<0.001	0.62	1.53	0.39
Quality of life – psychological health (WHOQOL-BREF)
Baseline	12.28	2.76	12.28	2.76	​	​	​	​	​
Week 8	12.17	3.25	11.84	2.91	0.30	0.172	−0.13	0.72	0.11
Week 24	12.43	3.37	11.87	3.11	0.48	0.029	0.05	0.92	0.17
Quality of life – social relationships (WHOQOL-BREF)
Baseline	13.04	3.61	13.05	3.21	​	​	​	​	​
Week 8	12.63	3.72	12.35	3.22	0.32	0.281	−0.26	0.91	0.09
Week 24	12.73	3.82	12.54	3.55	0.21	0.483	−0.38	0.81	0.06
Quality of life – environment (WHOQOL-BREF)
Baseline	15.21	2.54	14.90	2.36	​	​	​	​	​
Week 8	15.25	2.49	14.67	2.22	0.22	0.253	−0.16	0.61	0.09
Week 24	15.25	2.69	14.82	2.50	0.04	0.855	−0.35	0.43	0.02
Psychological need satisfaction and frustration – autonomy satisfaction (BPNSFS)
Baseline	13.02	3.09	13.11	3.36	​	​	​	​	​
Week 8	12.79	3.41	12.68	3.24	0.13	0.660	−0.46	0.73	0.04
Week 24	12.93	3.11	12.55	3.61	0.36	0.242	−0.25	0.97	0.11
Psychological need satisfaction and frustration – autonomy frustration (BPNSFS)
Baseline	12.12	4.04	12.28	3.93	​	​	​	​	​
Week 8	11.69	4.08	12.85	4.10	−0.90	0.011	−1.58	−0.21	−0.23
Week 24	12.07	4.31	12.93	4.03	−0.69	0.053	−1.39	0.01	−0.17
Psychological need satisfaction and frustration – relatedness satisfaction (BPNSFS)
Baseline	16.43	3.19	17.07	2.82	​	​	​	​	​
Week 8	15.84	3.68	16.36	3.14	0.04	0.902	−0.56	0.63	0.01
Week 24	16.16	3.18	16.27	3.68	0.45	0.142	−0.15	1.06	0.15
Psychological need satisfaction and frustration – relatedness frustration (BPNSFS)
Baseline	7.17	3.23	7.20	3.26	​	​	​	​	​
Week 8	8.04	3.59	7.38	3.45	0.67	0.019	0.11	1.23	0.21
Week 24	7.65	3.56	7.64	3.78	0.09	0.766	−0.48	0.66	0.03
Psychological need satisfaction and frustration – competence satisfaction (BPNSFS)
Baseline	14.78	3.02	14.45	3.53	​	​	​	​	​
Week 8	14.36	3.52	14.27	3.43	−0.35	0.256	−0.95	0.25	−0.11
Week 24	14.45	3.54	14.16	3.55	−0.12	0.690	−0.73	0.49	−0.04
Psychological need satisfaction and frustration – competence frustration (BPNSFS)
Baseline	9.08	3.63	8.63	3.87	​	​	​	​	​
Week 8	8.58	3.60	9.21	4.00	−0.94	0.002	−1.53	−0.35	−0.25
Week 24	8.67	3.77	9.12	4.20	−0.75	0.014	−1.35	−0.15	−0.20
Motivational incongruence – approach goals (K-INK)
Baseline	3.18	0.69	3.10	0.76	​	​	​	​	​
Week 8	3.18	0.85	2.98	0.83	0.10	0.124	−0.03	0.23	0.14
Week 24	3.15	0.92	2.95	0.93	0.09	0.168	−0.04	0.23	0.12
Motivational incongruence – avoidance goals (K-INK)
Baseline	2.31	0.74	2.43	0.75	​	​	​	​	​
Week 8	2.17	0.76	2.46	0.86	−0.18	0.027	−0.34	−0.02	−0.24
Week 24	2.27	0.87	2.38	0.95	−0.01	0.864	−0.18	0.15	−0.01

WHO-5, World Health Organization-Five Well-being Index (range: 0–25); WHOQOL-BREF, World Health Organization Quality of Life Questionnaire-brief version (range: 4–20); BPNSFS, Basic Psychological Need Satisfaction and Frustration Scale (range: 1–20); K-INK, Incongruence Questionnaire; approach goals range: 1–5; avoidance goals range: 1–5; Diff_adj, adjusted mean difference derived from linear mixed model; 95% CI, 95% confidence interval of the adjusted mean difference; ES, effect size (Cohen’s d); M, unadjusted means; SD, standard deviation.

Significant p values are displayed in bold.

Secondary Outcomes Relating to Co-Occurrence of Comorbidities

Depression

Between-group comparisons showed small treatment effects in favor of the digital CBT-I group after 8 (p < 0.001, d = −0.43) and 24 weeks (p < 0.001, d = −0.45) post-randomization, suggesting a reduction in depressive symptoms due to the digital CBT-I intervention.

Anxiety

The results of the linear mixed model showed small ESs at 8 (p < 0.001, d = −0.27) and 24 weeks (p < 0.001, d = −0.40) post-randomization. The dCBT-I group showed reduced anxiety symptoms compared to the WLC. See Table 6 for statistical analyses of all secondary outcomes relating to co-occurrence of those comorbidities.

Table 6.

Between-group effects of dCBT-I versus control group on secondary outcomes relating to co-occurrence of comorbidities

​	dCBT-I		Control group		Diffadj	p value	95% CI		ES
	M	SD	M	SD
Depression (ADS-K)
Baseline	19.46	8.29	19.03	7.91	​	​	​	​	​
Week 8	16.75	9.22	20.07	8.37	−3.52	<0.001	−5.10	−1.95	−0.43
Week 24	16.66	10.14	20.17	9.05	−3.63	<0.001	−5.21	−2.05	−0.45
Anxiety (STAI-T)
Baseline	48.90	10.81	48.92	10.26	​	​	​	​	​
Week 8	47.18	12.28	50.47	10.83	−2.89	<0.001	−4.54	−1.25	−0.27
Week 24	46.22	12.30	50.97	11.05	−4.26	<0.001	−5.91	−2.61	−0.40

ADS-K, Allgemeine Depressionsskala Kurzversion (Center for Epidemiological Studies Depression Scale), range: 0–45; STAI-T, Trait Version of the State-Trait Anxiety Inventory, range: 20–80; Diff_adj, adjusted mean difference derived from linear mixed model; 95% CI, 95% confidence interval of the adjusted mean difference; ES, effect size (Cohen’s d); M, unadjusted means; SD, standard deviation.

Significant p values are displayed in bold.

Secondary Outcomes Relating to Dreams and Nightmares

Dream Recall Frequency

No treatment effects were found regarding dream recall frequency between the dCBT-I group and the WLC after either 8 (p = 0.147) or 24 weeks (p = 0.845) post-randomization.

Nightmare Frequency

There were also no treatment effects between the dCBT-I group and the WLC in terms of nightmare frequency at 8 (p = 0.732) or 24 weeks (p = 0.553) post-randomization.

Nightmare Distress

Overall, the dCBT-I group showed a reduced nightmare distress compared to the WLC with small ESs both at 8 (p < 0.001, d = −0.29) and 24 weeks (p < 0.001, d = −0.30) post-randomization in the total score of the NDQ. The superiority of the dCBT-I group compared to the WLC was also demonstrated in terms of the “general nightmare distress” subscale with small ESs at 8 (p < 0.001, d = −0.28) and 24 weeks (p = 0.002, d = −0.26) post-randomization. A similar result was obtained for the “Impact on Sleep” subscale, where the dCBT-I group demonstrated reduced impact on sleep compared to the WLC with small ESs at 8 (p = 0.004, d = −0.28) and 24 weeks (p < 0.001, d = −0.33) post-randomization. In relation to the subscale “Impact on Reality Perception,” no superiority of the dCBT-I group could be identified at either time point (p = 0.217 and p = 0.111). See Table 7 for statistical analyses of all secondary outcomes relating to dreams and nightmares.

Table 7.

Between-group effects of dCBT-I versus control group on secondary outcomes relating to dreams and nightmares

​	dCBT-I		Control group		Diffadj	p value	95% CI		ES
	M	SD	M	SD
Dream recall frequencya
Baseline	13.45	20.81	11.79	17.25	​	​	​	​	​
Week 8	9.03	9.50	11.34	13.94	−2.16	0.147	−5.09	0.76	−0.11
Week 24	9.42	8.61	11.14	12.09	−0.30	0.845	−3.28	2.69	−0.02
Nightmare frequencya
Baseline	4.34	9.58	2.98	11.10	​	​	​	​	​
Week 8	2.83	9.15	3.37	9.00	0.25	0.732	−1.18	1.68	0.02
Week 24	3.16	7.12	3.05	8.05	0.44	0.553	−1.02	1.89	0.04
Nightmare distress total score (NDQ)
Baseline	29.43	9.76	26.91	9.86	​	​	​	​	​
Week 8	26.79	10.52	28.30	10.65	−2.81	<0.001	−4.43	−1.19	−0.29
Week 24	26.44	10.80	27.47	10.71	−2.94	<0.001	−4.58	−1.30	−0.30
General nightmare distress (NDQ)
Baseline	13.15	5.71	11.90	5.35	​	​	​	​	​
Week 8	11.78	5.83	12.47	5.73	−1.57	<0.001	−2.47	−0.67	−0.28
Week 24	11.70	6.06	12.23	5.86	−1.45	0.002	−2.36	−0.53	−0.26
Impact on sleep (NDQ)
Baseline	8.23	2.68	7.47	2.66	​	​	​	​	​
Week 8	7.39	2.84	7.80	2.79	−0.76	0.004	−1.27	−0.24	−0.28
Week 24	7.16	2.67	7.45	2.89	−0.88	<0.001	−1.40	−0.36	−0.33
Impact on reality perception (NDQ)
Baseline	8.06	2.82	7.55	2.88	​	​	​	​	​
Week 8	7.62	2.91	8.04	3.15	−0.38	0.217	−0.98	0.22	−0.13
Week 24	7.57	3.13	7.78	3.12	−0.50	0.111	−1.10	0.11	−0.18

NDQ, Nightmare Distress Questionnaire, total score range: 0–65; general nightmare distress range: 0–30; impact on sleep range: 0–15; impact on reality perception range: 0–20; Diff_adj, adjusted mean difference derived from linear mixed model; 95% CI, 95% confidence interval of the adjusted mean difference; ES, effect size (Cohen’s d); M, unadjusted means; SD, standard deviation.

Significant p values are displayed in bold.

^aWas assessed using an open-response format.

Adherence, Treatment Satisfaction, and Adverse Events

A total of 73.79% of participants (n = 76) completed at least half of the 10 core modules, while 38.81% (n = 41) progressed to complete the final module. Among those who received dCBT-I, 51 participants (49.51%) reported being satisfied or very satisfied with the intervention. Additionally, 22 participants (21.4%) expressed a neutral attitude regarding their satisfaction. A total of 38 participants (36.89%) indicated that their expectations of dCBT-I were largely or completely met, whereas 27 participants (26.2%) felt their expectations were at least partially fulfilled. A small minority (n = 8, 7.8%) reported that the intervention did not meet their expectations at all, primarily citing the lack of specific support for chronic pain as the main reason. Furthermore, over half of the participants (n = 64, 62.14%) stated that they had largely or completely engaged conscientiously with the intervention, and only 1 participant (1%) admitted to not completing the modules conscientiously. Importantly, no adverse events were reported by participants in either group throughout the study duration.

Discussion

Interpretation of Findings

The primary aim of this study was to investigate whether the implementation of dCBT-I, in addition to regular care and compared to a WLC, can lead to a reduction in insomnia severity in a sample of participants with chronic pain and insomnia. Additional treatment with dCBT-I was shown to be superior to the WLC, which received only regular care, with large ESs at all time points (ds = −1.18 and −1.32). These effects are within the range of, or even exceed, previous research findings on the effectiveness of dCBT-I in individuals with insomnia and chronic pain, both in face-to-face settings (SMD = −0.99 [32]) and digital interventions (d = −0.48 [46]). They are also comparable to ESs reported in meta-analyses evaluating dCBT-I effectiveness [42, 43, 93]. Responder and remission rates after 8 weeks (31.07% and 19.42%, respectively) were lower than those observed in studies that did not include additional comorbidities as an inclusion criterion [48, 94]. This may suggest that the presence of two coexisting disorders makes symptom reduction or remission more challenging to achieve.

The observed improvements in all secondary sleep-related outcomes, including reductions in daytime sleepiness, fatigue, and dysfunctional beliefs and attitudes about sleep, underscore the multiple benefits of dCBT-I. Positive effects were also found for well-being and the physical and, to some extent, psychological components of quality of life. Additionally, partial reductions were observed in frustration of psychological needs and motivational avoidance goals.

Regarding nightmares, a significant decrease in nightmare distress was noted in the dCBT-I group, despite no change in nightmare frequency. These findings suggest that dCBT-I not only targets the core symptoms of insomnia but also positively influences broader cognitive and physiological processes associated with sleep disorders. Nightmare distress has been linked to elevated presleep arousal, rumination, and disrupted emotional processing during sleep; see, for example, the affect network dysfunction model of nightmares [95]. In this context, recent research emphasizes the role of sleep in attenuating emotional and physiological arousal at night, thereby supporting emotional regulation and resilience during the day [96].

Components of dCBT-I, such as cognitive restructuring and behavioral sleep stabilization, may help reduce these factors and thereby alleviate nightmare symptoms. However, as these mechanisms were not directly assessed in the present study, this interpretation remains speculative and should be addressed in future research. The comprehensive effects highlight dCBT-I’s potential as an integrative intervention that addresses both the immediate symptoms and downstream consequences of sleep disorders, even in a highly burdened sample.

The results related to pain-related symptoms are generally heterogeneous. It is important to recognize that pain is a multidimensional construct encompassing sensory, affective, and cognitive components. The sensory dimension refers to the perception of pain itself, the affective component focuses on the emotional experience of pain, and the cognitive dimension includes the evaluation of pain and the subsequent reactions to it [97]. Research on brain activity patterns suggests that in chronic pain, the affective dimension becomes increasingly dominant, while the sensory dimension diminishes in importance [88]. In this context, the contradictory findings regarding the sensory component of pain intensity in our study are unsurprising, as participants had experienced pain for an average of 12 years. Such a prolonged history likely reflects a high degree of chronicity, in which the sensory aspect plays a subordinate role. Chronic pain, deeply integrated into self-perception, is often accompanied by significant emotional distress as well as cognitive and physical impairments [21]. As both cognitive [21] and affective [98] components gain prominence in the chronification process, it is plausible that the observed reduction in pain-related impairment and the increase in perceived life control following dCBT-I are linked to these modifiable dimensions. By addressing these factors through targeted sleep improvements, dCBT-I may help reduce the overall degree of pain chronification.

Interestingly, no significant effect was observed for affective distress. However, a descriptive trend favoring dCBT-I was evident, suggesting that the effect may be very small and could not be detected with the current sample size and statistical power.

Limitations

While this study provides valuable insights into the effectiveness of dCBT-I in patients with comorbid chronic pain, several limitations must be acknowledged. The sample included individuals with various types of chronic pain, which differ in their association with insomnia. For instance, back, limb, head, and joint pain are strongly linked to insomnia symptoms, comparable to affective disorders, whereas abdominal or stomach pain shows weaker associations [99]. This heterogeneity may have influenced the results. The gender distribution was notably skewed, with 88% of participants being women. Although this reflects the higher prevalence of insomnia [100], chronic pain [101], and their co-occurrence [14] in women, it limits the generalizability of the findings to male patients. The exclusive use of online recruitment channels may have attracted individuals more familiar with digital media, introducing selection bias and limiting representativeness for populations with less access to or experience with digital tools. Moreover, the analyses did not control for the influence of concurrent treatments during the intervention period. Although such treatments were documented descriptively, changes in medical or psychological co-interventions were not systematically assessed, reducing internal validity and limiting causal conclusions. The use of a WLC group with access to usual care may also have led to an overestimation of ESs due to the absence of an active comparator. Nevertheless, previous studies have shown that dCBT-I maintains its effectiveness even in comparison with active control conditions [49, 102, 103], suggesting the observed effects are not solely attributable to nonspecific factors. Finally, the open-label design may have introduced expectation-related bias, as participants were aware of their group allocation, potentially influencing their perception of treatment effectiveness [104].

Where Do We Go from Here?

The present results indicate that dCBT-I shows effectiveness comparable to, and to some extent potentially exceeding, that of face-to-face CBT-I in patients with comorbid insomnia and chronic pain [31, 32, 36]. While this highlights the potential of fully automated interventions, their integration into primary care warrants further investigation. Practical barriers such as limited motivation, technical challenges, and data privacy concerns may limit uptake in primary care settings [105]. At the same time, advantages like scalability, cost-effectiveness, and flexible access underscore the relevance of digital formats for broader dissemination. These considerations point to the need for continued research on the feasibility, acceptability, and clinical effectiveness of dCBT-I in diverse healthcare systems.

Although the current study provides promising evidence from a large and heterogeneous sample, there remains a theory-practice gap in the mechanistic understanding of the treatment (i.e., how does CBT-I reduce pain symptoms?). Considering that the explanatory power is also hindered by the presence of multiple therapeutic components, a dismantling trial design may be warranted to prospectively investigate the contribution of specific treatment components [106]. In line with the framework of the Research Domain Criteria (RDoC) [107], such a mechanism-dedicated trial should also incorporate biological and neurophysiological markers to investigate hypothesized mechanisms more directly, such as neuroinflammatory processes (e.g., blood levels of IL-6, TNF-α, CRP), neural activation patterns during pain processing (e.g., fMRI), and central pain modulation (e.g., conditioned pain modulation paradigms).

The optimal sequencing of treatments for comorbid insomnia and chronic pain also remains an open question. Both conditions are mutually reinforcing and may benefit from integrated or sequential approaches. Due to its transdiagnostic impact and modifiable nature, dCBT-I may represent a pragmatic first-line treatment within stepped-care models. However, head-to-head comparisons of different sequencing strategies, such as insomnia-first, pain-first, or parallel interventions, are needed to guide evidence-based decision-making in clinical care.

Conclusion

Overall, this study demonstrates the high effectiveness of dCBT-I in reducing insomnia severity and improving sleep-related outcomes in individuals with chronic pain. The broad-ranging benefits observed underscore the intervention’s potential in addressing complex symptomatology. While sensory pain-related outcomes were heterogeneous, significant improvements in pain interference and perceived life control suggest that dCBT-I may target shared modifiable mechanisms of chronic pain and insomnia. These include maladaptive cognitions, avoidance behavior, and reduced self-efficacy, core components in theoretical models such as the fear-avoidance model [19, 20] and the triple network model [21]. A recent cognitive model further specifies how overlapping cognitive biases, such as hypervigilance, negative interpretation of somatic sensations, and selective attention to pain- or sleep-related threats, may mutually reinforce both conditions [22]. dCBT-I may effectively disrupt these transdiagnostic patterns. Given the existing shortfall in psychotherapy availability [108], the digital implementation of CBT-I presents an innovative approach to closing care gaps. Beyond preventing the escalation of chronic conditions, it may also enhance patients’ overall functioning and well-being. To build on these findings, future research should focus on replicating results in primary care settings, exploring long-term outcomes, and identifying the most effective components, mechanisms of change and delivery formats of dCBT-I for individuals with comorbid insomnia and chronic pain.

Acknowledgments

We thank Alessandra Russo, Klara Sophie van Waasen, Lara Kristin Haupt, Lukas Uhlich, Marie-Therese Steil, and Nina Platzer for their dedicated help during data collection.

Statement of Ethics

This study protocol was reviewed and approved by Ethics Committee of Heinrich Heine University Duesseldorf, Approval No. GI01_2022_01, and preregistered at the German Clinical Trials Register (DRKS00030972). Written informed consent was obtained from all participants in digital form both before participating in the online screening and before participating in the baseline assessment.

Conflict of Interest Statement

At the time of study planning and implementation, J.S. was a part-time employee of mementor DE GmbH, a company that specializes in the digital delivery of cognitive behavioral therapy for sleep improvement (somnio). L.F.M. is a salaried employee of mementor DE GmbH. A.G. declares nonfinancial support in the form of no cost access to somnio for use in clinical research.

Funding Sources

This study was sponsored by mementor DE GmbH, the company that developed the dCBT-I somnio which is used in this study. The corresponding author designed and conducted the study as part of her doctoral thesis in compliance with good scientific practice, and it was contractually stipulated that the sponsor has no influence on the design, data collection, data analysis, and reporting of this study.

Author Contributions

Jennifer Schuffelen: conceptualization, project administration, methodology, formal analysis, visualization, writing – original draft, review, and editing, and validation. Leonie F. Maurer and Annika Gieselmann: conceptualization, methodology, validation, writing – review and editing, and supervision.

Funding Statement

This study was sponsored by mementor DE GmbH, the company that developed the dCBT-I somnio which is used in this study. The corresponding author designed and conducted the study as part of her doctoral thesis in compliance with good scientific practice, and it was contractually stipulated that the sponsor has no influence on the design, data collection, data analysis, and reporting of this study.

Data Availability Statement

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author J.S. upon reasonable request.