Effect of Cognitive Behavioral Therapy for Insomnia in Patients With Co-Morbid Insomnia and Sleep Apnea: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

Search Strategy

Computerized systematic literature data searches were conducted in PubMed, Scopus, EBSCO, Taylor and Francis, Web of Science, and Sage. The following keywords were used: “Obstructive Sleep Apnea,” “Insomnia,” “Cognitive Behavioral Therapy,” and “Randomized.” We simultaneously added several Medical Subject Headings (MeSH) terms as well as other free text terms to construct database-specific search terms. The search details of each database can be seen in Supplementary Table 1 (in the online-only Data Supplement).

Study Selection

Search results from each database were collected and organized collectively using Rayyan.ai software. After deduplication, the remaining articles were screened by title and abstract. All articles were then pooled for screening based on eligibility criteria. The rationale behind the exclusion of each article at each stage was stated in the accompanying PRISMA diagram. The entire literature search and study selection process was carried out by DFFA, ZNS, and MFH. Any disagreements were resolved through joint discussions. Details of the selection process can be seen as a PRISMA flowchart in Fig. 1.

Eligibility Criteria

We applied the Population, Intervention, Comparison, Outcome framework in formulating the eligibility criteria. To be included in systematic reviews and meta-analysis, studies had to meet the following criteria: 1) study population consisted of patients with COMISA; 2) involved CBT-I as the main therapy; 3) RCT study; 4) English language study. Studies will be excluded if: 1) studies only available as abstracts; 2) studies with incomplete data; 3) review or protocol studies.

Data Extraction and Quality Assessment

The four authors (DFFA, ZNS, and MFH) extracted data from each study, starting with data tabulation in a spreadsheet by MFH, after which DFFA examined the data collected for eligibility. Data extracted included the first author’s name and year of publication, study location, subject, route of administration, dose of administration, duration of treatment, treatment outcome, and p-value of each study outcome. The results of data extraction from each study were presented qualitatively and quantitatively in tabular format. The pooled studies were further reviewed for methodological quality by two independent reviewers (DFFA and ZNS) using the Cochrane risk-of-bias tool for randomized trials version 2 (RoB 2) [18]. Discrepancies in decisions were then resolved with the involvement of a third reviewer (MFH). The results of the risk of bias assessment are then visualized using robvis [13].

Data Synthesis and Statistical Analysis

Primary quantitative analysis was performed using Review Manager 5.4.1 (Cochrane Collaboration). Outcomes were analyzed with a fixed effect model if the level of heterogeneity (I²) was below 50% and with a random effect model if the level of heterogeneity (I²) was greater than 50%. For effect size, the mean difference (MD) was used, with a 95% confidence interval (CI) considered that a significant p-value was less than 5% (p<0.05). For all meta-analysis, forest plots were presented. Heterogeneity was investigated using Higgins I² value, which indicates heterogeneity can be classified as negligible (<25%), low (25%–49%), moderate (50%–74%), or high (≥75%) [18]. If there are uncertain data reporting results, such as the absence of standard deviation (SD) (change from baseline value), because such information is often not available in the experiment report and must be inputted again [19]. Missing values were calculated from similar trials that were included in the same meta-analysis. In this case, we first calculated the correlation using the following formula: Corr=SD²_baseline+SD²_follow-up-SD²_change/2×SD_BASELINE×SD_follow-up. The SD of change from the base value is then calculated using the formula: SD²_change=√SD 2baseline+SD²_follow-up-(2×Corr×SD_baseline×SD_follow-up).

The results of the SD values that have been obtained are then included along with the mean of each data from each study in the extraction table. The mean and SD change values included in Supplementary Tables 2-6 (in the online-only Data Supplement) are the final results taken from each inclusion study.

Study Selection and Quality Assessment

From the mentioned database, 687 articles were retrieved. After removing 132 duplicates and screening titles/abstracts, 23 potential articles were selected with one article being left out due to an irretrievable paper. After full text review, ten studies consisting of Bensen-Boakes et al. [20] in 2022, Tu et al. [21] in 2022, Alessi et al. [22] in 2021, Sweetman et al. [23] in 2021, Ong et al. [24] in 2020, Sweetman et al. [25] in 2020, Sweetman et al. [26] in 2020, Sweetman et al. [27] in 2019, Fung et al. [28] in 2016, and Richards et al. [29] in 2007 were included in the systematic review and meta-analysis. The study selection process is detailed in the PRISMA flowchart (Fig. 1). A total of ten studies were then quality reviewed using the RoB 2 as shown in the traffic light plot (Fig. 2) and summary (Fig. 3).

Fig. 2.

Risk of Bias 2 (RoB2) assessment traffic light plot.

Fig. 3.

Risk of Bias 2 (RoB2) assessment summary.

Study Characteristics

This meta-analysis and review used ten RCT studies as its source of data and information. Published between 2007 and 2022, these studies compared the effect of CBT-I versus control or PAP intervention for COMISA patients. The complete study characteristics can be seen in Table 1.

Table 1.

Study characteristics

Effect of CBT-I vs. Control on Sleep Diary Outcomes

The sleep parameter reviewed in this analysis is sleep quality as measured by sleep diary output. There are a total of five studies that compare the results of sleep diary output in patients given CBT-I intervention and controls, namely in the studies of Fung et al. [28] in 2016, Sweetman et al. [25] in 2020, Sweetman et al. [26] in 2020, Bensen-Boakes et al. [20] in 2022, Sweetman et al. [23] in 2021. In all five studies, there was a change in the sleep efficiency (SE) value in patients who were given the CBT-I intervention when compared to the control. Based on Fig. 4A, in the studies of Fung et al. [28] in 2016, and Bensen-Boakes et al. [20] in 2022, a meta-analysis was conducted and a significant increase in the SE value was found (MD=7.54, 95% CI=4.07 to 11.00, p<0.001). The same thing was explained in the Sweetman et al.’s study [23] (2021) where in that study a significant increase in the SE parameter was found (p<0.001). Meanwhile, Sweetman et al. [25] in 2020 found an insignificant increase (p>0.08). In addition, based on Fig. 4B, overall, there was also a significant change in sleep diary output aspects for sleep onset latency (SOL) and wake after sleep onset (WASO) in the CBT-I intervention compared to the control (MD=-19.97, 95% CI=-22.26 to -17.68, p<0.001). Subgroup analysis was also performed on each outcome with SOL scores significantly lower in the CBT-I intervention compared to control (MD=-18.29, 95% CI=-27.98 to -8.60, p<0.001). WASO scores were also found to be significantly lower in the CBT-I intervention compared to the control (MD=-19.20, 95% CI=-19.75 to -18.64, p<0.001). Similar results were found in the Sweetman et al.[25] in 2020 and Sweetman et al. [26] in 2020 studies with a significant decrease in WASO parameters (p=0.016 and p=0.031, respectively).

Fig. 4.

A: Meta-analysis of the effect of CBT-I vs. control on sleep diary–ascending outcomes. B: Meta-analysis of the effect of CBT-I vs. control on sleep diary–descending outcomes. CBT-I, cognitive behavioral therapy for insomnia; SE, sleep efficiency; CI, confidence interval; SD, standard deviation; SOL, sleep onset latency; WASO, wake after sleep onset; IV, inverse variance.

Effect of CBT-I vs. Control on Sleep Questionnaire Outcomes

Two studies, Sweetman et al. [23] (2021), and Sweetman et al. [25] (2020), compared sleep questionnaire outcomes in CBT-I interventions compared to controls. In both studies, there was a significant decrease in ESS questionnaire scores (p=0.005 and p=0.031, respectively). Furthermore, Sweetman et al.’s study [23] (2021) also showed a significant decrease in Dysfunctional Beliefs and Attitudes about Sleep (DBAS) and Insomnia Severity Index (ISI) scores (p=0.018 and p≤0.003, respectively).

Effect of CBT-I+PAP vs. PAP on Sleep Diary Outcomes

There were a total of three studies that compared sleep diary outcomes in patients with CBT-I+PAP and PAP interventions [21,22,27]. Based on Fig. 5A, overall, there was a non-significant change in sleep diary outcome for both parameters (total sleep time [TST] and SE) in the CBT-I+PAP intervention compared to PAP (MD=0.81, 95% CI: -4.39 to 6.01, p=0.76). Subgroup analysis was conducted on these two parameters. The TST value significantly increased in CBT-I+PAP intervention compared to PAP (MD=-5.47, 95% CI=-9.45 to -1.48, p=0.007) with the TST value more inclined towards PAP intervention. SE scores also increased significantly in the CBT-I+PAP intervention compared to PAP (MD=4.19, 95% CI=2.93 to 5.45, p<0.001). Whereas, in the sleep diary outputs that decreased based on Fig. 5B (SOL, WASO, and time in bed [TIB]), overall, there was a significant change in the CBT-I+PAP intervention compared to PAP (MD= -16.92, 95% CI=-27.08 to -6.67, p<0.001). Subgroup analysis was then conducted to examine each parameter in the intervention. SOL and TIB parameter values decreased significantly (MD=-12.51, 95% CI=-13.01 to -12.00, p<0.001; MD=-31.91, 95% CI=-37.81 to -26.00, p<0.001, respectively). In WASO parameter values, there was a decrease but not significant in CBT-I+PAP intervention compared to PAP (MD=-7.06, 95% CI=-14.33 to 0.21, p=0.06).

Fig. 5.

A: Meta-analysis of effect of CBT-I+PAP vs. PAP on sleep diary outcomes–ascending outcomes. B: Meta-analysis of effect of CBTI+ PAP vs. PAP on sleep diary outcomes–descending outcomes. CBT-I, cognitive behavioral therapy for insomnia; PAP, positive airway pressure; TST, total sleep time; SE, sleep efficiency; CI, confidence interval; SD, standard deviation; SOL, sleep onset latency; WASO, wake after sleep onset; TIB, time in bed; IV, inverse variance.

Effect of CBT-I+PAP vs. PAP on Sleep Actigraphy

There are two studies that compare the results of sleep actigraphy in CBT-I+PAP and PAP interventions, namely in Alessi et al. [22] in 2021 and Tu et al. [21] in 2022. Based on Fig. 6, in the SE parameter through sleep actigraphy, there was an insignificant improvement in the CBT-I+PAP intervention compared to PAP (MD=0.08, 95% CI=-4.99 to 4.83, p=0.97).

Fig. 6.

Meta-analysis of effect of CBT-I+PAP vs. PAP on sleep actigraphy. CBT-I, cognitive behavioral therapy for insomnia; PAP, positive airway pressure; SE, sleep efficiency; CI, confidence interval; SD, standard deviation; IV, inverse variance.

Effect of CBT-I+PAP vs. PAP on Sleep Questionnaires

Sleep questionnaires that were involved consist of Functional Outcomes of Sleep Questionnaire (FOSQ), Pittsburgh Sleep Quality Index (PSQI), Epworth Sleepiness Scale (ESS), DBAS, ISI, and Flinders Fatigue Scale (FFS). In total, four studies compared sleep questionnaire scores in CBT-I+PAP and PAP interventions [21,22,24,27]. Based on Fig. 7A, in the parameters that increased (FOSQ), there was a significant change in CBT-I+PAP and PAP interventions (MD=2.13, 95% CI=1.88 to 2.39, p<0.001). Whereas, in the parameters that decreased based on Fig. 7B (PSQI, ESS, DBAS, ISI, and FFS), overall, there was a significant change in questionnaire score values in CBT-I+PAP and PAP interventions (MD=-5.91, 95% CI=-10.41 to -1.40, p=0.01). Subgroup analysis was performed for PSQI, ESS, DBAS, ISI, and FFS questionnaire scores. There was a significant decrease in PSQI, DBAS, ISI, and FFS questionnaire scores in CBT-I+PAP and PAP interventions (MD=-1.01, 95% CI=-1.94 to -0.07, p=0.03; MD=-19.11, 95% CI=-34.66 to -3.55, p=0.02; MD=-3.73, 95% CI=-4.92 to -2.53, p<0.001; MD=-3.16, 95% CI=-4.70 to -1.61, p<0.001, respectively). ESS scores decreased insignificantly in CBT-I+PAP and PAP interventions (MD=-0.76, 95% CI=-1.84 to 0.31, p=0.16).

Fig. 7.

A: Meta-analysis of effect of CBT-I+PAP vs. PAP on sleep questionnaire–ascending outcomes. B: Meta-analysis of effect of CBTI+ PAP vs. PAP on sleep questionnaire–descending outcomes. CBT-I, cognitive behavioral therapy for insomnia; PAP, positive airway pressure; FOSQ, Functional Outcomes of Sleep Questionnaire; PSQI, Pittsburgh Sleep Quality Index; ESS, Epworth Sleepiness Scale; DBAS, Dysfunctional Beliefs and Attitudes about Sleep; ISI, Insomnia Severity Index; FFS, Flinders Fatigue Scale; CI, confidence interval; SD, standard deviation; IV, inverse variance.

Effect of CBT-I+PAP vs. PAP on PAP Adherence

There are three studies that compare PAP adherence parameters in CBT-I+PAP and PAP interventions, namely Richards et al. [29] in 2007, Ong et al. [24] in 2020, and Alessi et al. [22] in 2021. Based on Fig. 8, overall, there was a non-significant change in PAP adherence parameters in the CBT-I+PAP intervention compared to PAP (MD=-0.28, 95% CI=-0.71 to 0.15, p=0.20). The Ong et al.’s study [24] (2020) also showed insignificant results in the change of PAP adherence parameters (p=0.4852). However, in this parameter, the Alessi et al.’s study [22] (2021) had a trend towards the provision of CBT-I+PAP interventions with significant changes in PAP compliance values.

Fig. 8.

Meta-analysis of the effect of CBT-I+PAP vs. PAP on PAP adherence. CBT-I, cognitive behavioral therapy for insomnia; PAP, positive airway pressure; CI, confidence interval; SD, standard deviation; IV, inverse variance.

Effects of CBT-I on COMISA Symptoms

The main finding of this systematic review and meta-analysis was the association between CBT-I and major improvements in the severity of insomnia in patients with COMISA through various parameters including sleep diary, sleep actigraphy, sleep questionnaire, and PAP compliance. CBT-I improved insomnia when given separately from PAP therapy or when given together with PAP therapy. These findings were based on 10 studies involving 1113 patients with COMISA, including seven that were subjected to further meta-analysis. Through the entire series of data processing and analysis that has been carried out, five of the ten RCTs involved in this study compared the provision of CBT-I therapy with control. Meanwhile, five of the ten studies compared the combination of CBT-I and PAP therapy with PAP alone. Both comparisons reported better insomnia improvement in participants with therapy involving CBT-I, either CBT-I alone or in combination with PAP. Based on this data, it can be suggested that patients with COMISA be treated with CBT-I as an adjunct therapy to COMISA management which generally only targets the disease condition from the OSA point of view or, in other words, only uses PAP therapy. Several previously conducted meta-analyses have also shown that CBT-I is an insomnia treatment that has a high potential for therapeutic success in treating cooccurring physical and mental health problems [14,30-33], which in this case occurs in patients with COMISA. This is thought to be one of the results of the direct role of CBT-I on insomnia, which is the main symptom in patients with COMISA along with OSA. CBT-I specifically targets the psychological and behavioral factors that underlie the conditions for the perpetuation of insomnia [34]. However, in the future, until data from further research is found regarding the effect of CBT-I alone on COMISA, it is still recommended that treatment is based not only on mental health problems, namely insomnia, but also on physical health problems that occur in COMISA patients, namely OSA so that the combination of CBT-I with PAP should always be available. In view of this, with evidence that almost all patients with COMISA appear to respond to CBT-I, this intervention, combined with PAP, is a powerful pair of interventions that can treat the comorbid conditions of insomnia and OSA simultaneously.

Based on a meta-analysis conducted by Sweetman et al. [14] in 2023, in the process of adapting the use of CBT-I in patients with COMISA, several things must be considered. Some of these things such as the first, the effect of sleep restriction therapy, which is one of the components of CBT-I, can acutely result in increased daytime sleepiness. Second, the direction of stimulus control therapy and sleep restriction therapy needs to be modified in patients with COMISA who are already using PAP therapy where the time of use of PAP therapy has been previously prescribed. Third, the application of CBT-I as part of the therapy used for COMISA patients needs to be considered for use in patients with occupations as night-shift workers who do not allow patients to get sleep according to what has been prescribed, suggested, and directed to CBT-I therapy [35]. Therefore, further research needs to be done to customize and adapt CBT-I, not only limited to the night-shift worker population, but also to the entire input population of patients with COMISA.

The interplay between CBT-I and PAP therapy in managing COMISA invites a deeper exploration of theoretical frameworks underlying this synergy. At its core, CBT-I operates on the premise of targeting maladaptive cognitive and behavioral patterns that perpetuate insomnia [36]. This psychological reconditioning, when integrated with PAP therapy’s physiological intervention addressing OSA, exemplifies a biopsychosocial approach to complex comorbidities [37]. Such an approach inherently recognizes the multifactorial etiology of COMISA, where the intersection of psychological, behavioral, and physiological dimensions demands a multimodal therapeutic strategy [38].

The acute increase in daytime sleepiness resulting from sleep restriction therapy within CBT-I frameworks provides a critical lens through which to examine the temporal dynamics of therapeutic interventions. Theoretical considerations suggest that while sleep restriction serves to consolidate sleep and improve SE over time, its immediate effects may transiently disrupt homeostatic and circadian mechanisms [39]. This raises questions about the trade-offs between short-term dissonance and long-term gains, necessitating a model that predicts and mitigates these transitional challenges. Such a model might benefit from incorporating principles of chronobiology, which emphasize the temporal alignment of therapeutic interventions with the patient’s biological rhythms [40].

Additionally, the theoretical alignment between stimulus control therapy and PAP usage presents a compelling area for exploration. Both interventions share a foundational emphasis on consistency and habit formation—CBT-I seeks to regulate sleep-wake patterns through behavioral conditioning, while PAP adherence requires a similar behavioral commitment to nightly usage. Understanding this alignment through the lens of habit-formation theories could inform strategies to enhance patient compliance. For instance, applying principles from self-determination theory might elucidate how fostering intrinsic motivation in patients enhances adherence to both CBT-I and PAP protocols, creating a mutually reinforcing cycle of behavioral and physiological improvement [41].

The challenge of adapting CBT-I for specific populations, such as night-shift workers, further underscores the need for theoretical expansion. Night-shift work disrupts the circadian system, often leading to misalignment between internal biological clocks and external environmental cues [42]. Theoretical constructs such as phase-shift models could guide the adaptation of CBT-I, emphasizing the recalibration of sleep schedules in a manner that respects circadian constraints. Additionally, applying concepts from occupational health psychology could help contextualize CBT-I interventions within the broader psychosocial stressors experienced by shift workers, offering a more holistic understanding of their therapeutic needs [43].

Finally, the differential responses to CBT-I within the COMISA population invite theoretical inquiry into the mechanisms driving treatment efficacy. Cognitive-behavioral theory posits that the reappraisal of dysfunctional beliefs about sleep is central to CBT-I’s success [44]. However, in the context of COMISA, the presence of OSA adds an additional layer of complexity. The interaction between cognitive arousal, physiological arousal from apneic events, and the psychological burden of managing a chronic condition may require an expanded theoretical framework—one that integrates principles from stress and coping theories to address the compounded impact of dual pathologies.

In essence, the integration of CBT-I and PAP therapy for COMISA represents a theoretical convergence of psychological and physiological models of health. As research continues to elucidate this relationship, it becomes increasingly important to refine these theoretical underpinnings, ensuring they account for the diverse and complex experiences of patients. By doing so, the field can advance toward a more nuanced and adaptable framework for treating COMISA, one that bridges the gap between theory and practice with scientific rigor and clinical empathy.

Effect of CBT-I on PAP Use

According to Sweetman et al. [23] in 2021 patients with COMISA tend to have lower acceptance and adherence to PAP use when compared to OSA patients alone. Therefore, this study also directly reviews various existing RCT studies regarding the effect of CBT-I on PAP compliance in COMISA patients. As shown in Fig. 8, there was a change in the PAP adherence parameter. Although in this study, no significant overall change was found, however, in one of the RCT studies involved, namely Alessi et al. [22] in 2021, a significant increase in PAP adherence was found after the CBT-I intervention. Non-significant results were mentioned in the studies of Ong et al. [24] in 2020 and Richards et al. [29] in 2007. This difference in results could result from variations in sample size and the type of intervention provided in each of these studies.

Previously, in the study of Sweetman et al. [27] in 2019, the treatment applied to COMISA patients tended to only lean towards the provision of CPAP alone and was not combined with CBT-I on the basis that patients felt they received sufficient benefits from CPAP alone, and did not feel the need to use CBT-I interventions. However, through the results obtained in several studies, CBT-I can improve patient compliance in using PAP as a combination therapy. This method has also been suggested previously by a guideline for the management of chronic insomnia and OSA by Mysliwiec et al. [45] in 2020, which suggests providing PAP interventions that are preceded by treatment of the insomnia condition experienced, in this case using CBT-I therapy. This has been shown to increase patient compliance in using PAP as therapy for the COMISA condition experienced [22]. However, the underlying mechanism that causes the increase in PAP adherence due to the provision of CBT-I is still difficult to identify because in the Alessi et al.’s study [22] (2021), a combination of CBT-I and PAP interventions with motivational encouragement was used. Aside from the type of intervention used, the specific type of OSA that is involved will also fully influence the results that CBT-I therapy will have on PAP adherence. This is clearly seen where in the Alessi et al.’s study [22] (2021), the patients involved were patients with a minimum OSA classification of moderate OSA. Meanwhile, in the Ong et al.’s study [24] (2020), the patients involved were patients with mild, moderate, and severe OSA. This condition could also be one of the reasons why the results of the PAP adherence output in the two studies are quite different and as a result the overall results in Fig. 8 have a high Higgins I2 heterogeneity number along with the overall results are not significant.

The interplay between CBT-I and PAP adherence in COMISA patients exemplifies the intricate balance between psychological and physiological therapeutic paradigms. The variability observed across studies—ranging from significant improvements in adherence to non-significant findings—highlights the multifaceted nature of patient compliance. Theoretically, this variability may stem from the intersection of cognitive readiness, behavioral conditioning, and the biopsychosocial complexities unique to COMISA [46]. CBT-I, with its targeted focus on reshaping maladaptive sleep-related cognitions, provides a critical foundation for optimizing PAP adherence.

However, the mechanisms underlying this synergy remain an open question, necessitating further exploration into the dynamic relationship between insomnia alleviation and adherence behaviors.

One possible theoretical explanation lies in the concept of behavioral activation through cognitive restructuring. CBT-I addresses insomnia by reframing dysfunctional beliefs and establishing structured behavioral patterns conducive to restorative sleep [47]. This process may inadvertently enhance PAP adherence by fostering a sense of self-efficacy and control over sleep disturbances. In this context, adherence to PAP becomes a logical extension of the patient’s redefined relationship with sleep, reframed not as a passive dependency on a device but as an active commitment to holistic improvement. This aligns with self-determination theory, which posits that intrinsic motivation is pivotal for sustained behavior change, suggesting that CBT-I might operate as a catalyst for deeper engagement with PAP therapy [41].

The role of OSA severity as a moderating factor introduces a compelling layer of complexity to this interplay. The findings of Alessi et al. [22] in 2021, wherein patients with moderate or severe OSA demonstrated significant adherence improvements, suggest that CBT-I’s efficacy may be contingent upon the perceived salience of the sleep-disordered breathing condition. Theoretically, individuals with more severe OSA may experience greater symptom relief from PAP, reinforcing adherence through a feedback loop of perceived benefits [48]. Conversely, patients with mild OSA might lack this immediate reinforcement, potentially dampening the additive effects of CBT-I [49,50]. This raises the question of whether a tailored stratification approach—modulating CBT-I protocols based on OSA severity—could optimize adherence outcomes across the COMISA spectrum.

Timing and sequencing of interventions also emerge as critical theoretical considerations. Mysliwiec et al. [45] (2020) and Williams Buckley et al.’s [51] (2020) guideline advocating for the treatment of insomnia before initiating PAP aligns with transtheoretical models of behavior change, which emphasize the importance of readiness and precontemplation stages. By resolving insomnia as a primary barrier, CBT-I might pave the way for smoother transitions into PAP adherence, effectively addressing psychological resistance before introducing mechanical intervention. However, this sequential model warrants further theoretical inquiry: Could concurrent application of CBT-I and PAP harness synergistic effects, reinforcing adherence through simultaneous resolution of both insomnia and sleep apnea? Longitudinal studies exploring these paradigms would illuminate whether integration or sequencing yields superior patient outcomes [52].

The inclusion of motivational components within CBT-I, as highlighted in Alessi et al. [22] 2021, also deserves theoretical interrogation. Motivation may serve as a bridging mechanism between cognitive restructuring and behavior execution, addressing the practical and emotional barriers to PAP use. Drawing on principles of motivational interviewing, which emphasizes autonomy and self-directed change, future iterations of CBT-I could incorporate tailored motivational strategies to enhance engagement [53]. This integration would expand the scope of CBT-I from a strictly cognitive-behavioral model to a more holistic framework addressing the psychosocial dimensions of adherence.

Finally, the methodological heterogeneity observed across studies underscores the need for theoretical precision and standardization. The inconsistencies in sample characteristics, intervention designs, and outcome metrics obscure the nuanced effects of CBT-I on PAP adherence. A unified theoretical framework—grounded in principles of sleep science, behavioral psychology, and clinical epidemiology—could guide the development of standardized protocols, ensuring comparability and reproducibility. Furthermore, incorporating advanced statistical techniques such as meta-regression could elucidate the relative influence of moderating factors like OSA severity, demographic diversity, and baseline psychological profiles on adherence outcomes.

In conclusion, the integration of CBT-I with PAP adherence in COMISA patients is a microcosm of the broader challenge of aligning psychological and physiological interventions. By advancing theoretical models that capture the interplay of cognitive restructuring, behavioral activation, and motivational reinforcement, future research can refine this therapeutic paradigm. Such efforts will not only enhance adherence but also contribute to a deeper understanding of the complex biopsychosocial mechanisms underlying COMISA, ultimately improving patient outcomes and quality of life.

Strengths and Limitations

This study is the first systematic review and meta-analysis study to compare the outcomes of the measured parameters, namely sleep diary, sleep actigraphy, sleep questionnaire, and PAP adherence from RCT studies by comparing the provision of CBT-I alone or in combination as the beginning of PAP therapy, with controls or PAP intervention alone in patients with COMISA. However, in writing, there are several limitations.

First, this study only included ten RCT studies with a total participation of 1113 patients with COMISA, and only seven of them had sufficient data for meta-analysis. Although RCT studies have the highest evidence base value, the number of studies involved is still insufficient to be able to explain specifically how the effectiveness of providing CBT-I interventions, either given alone or combined with PAP, on each of the measured outcome parameters. In addition, the COMISA patient groups included in each study had different sample selection criteria, making it difficult for the authors to accurately predict the effect of CBT-I on the overall population. Therefore, further RCT studies in the future with a focus on each outcome and the same COMISA patient group selection criteria are expected to be conducted.

Second, this systematic review and meta-analysis study had quite high Higgins I² values in some of its outputs. This indicates that there is a high degree of heterogeneity in each of the studies included. This heterogeneity may arise, among others, as a result of the following: 1) The RCT studies involved had different COMISA patient selection criteria. For example, during the systematic review, we found several studies that examined cases where patient samples were drawn from a broad insomnia population and coincidentally contained people with COMISA. Most of these studies failed to reveal the impact of CBT-I among a subsample of COMISA participants.

2) Of the ten studies included, most involved COMISA patients with different disease classifications. This is evident from several studies that still involved patients with mild, moderate, and severe AHI grading.

3) There are differences in the number of CBT-I sessions provided in each study as well as the timing of the intervention which varies in the duration of time. This may contribute to the differences in intervention outcomes in the patient groups analyzed in each RCT study.

Third, the ten studies involved in this study were only conducted in the COMISA population in two countries, namely the United States and Australia. This results in a small sample coverage only in certain populations so that it still cannot explain the effect of CBT-I administration in patients with COMISA as a whole.

Recommendations

Based on the review of research limitations and literature review, there are several suggestions for future research in the form of:

1) Conduct a systematic review of the literature on the same topic using a larger sample of articles with similar sample selection criteria methods in the future.

2) A literature review of the literature on the same topic using more homogeneous data, both the disease classification of the samples involved, and the specifications of the interventions provided, so as to reduce the heterogeneity rate in the final results of the study analysis.

3) Involve studies conducted in various centers (multicenter) so that the generalization of the output results in systematic review studies and meta-analyses can be applied to various populations around the world.

Conclusion

Based on the results of our meta-analysis, CBT-I, both being used as a single therapy or as a combination with PAP, has a significant effect in improving insomnia and OSA conditions in patients with COMISA. Therefore, it can be concluded that the use of CBT-I in the ongoing therapies could be a promising treatment for COMISA. However, further studies that are more homogenous and thorough are needed to better understand the effectiveness of CBT-I towards patients with COMISA.