Skip to main content
NCBI home page
BMC Psychiatry logoLink to BMC Psychiatry
. 2025 Jan 28;25:77. doi: 10.1186/s12888-025-06532-w

Depressive, anxiety, and sleep disturbance symptoms in patients with obstructive sleep apnea: a network analysis perspective

Xue Luo 1,2,#, Shuangyan Li 1,2,#, Qianyun Wu 1,2, Yan Xu 1,2, Ruichen Fang 1,2, Yihong Cheng 1,2, Bin Zhang 1,2,✉
PMCID: PMC11773896  PMID: 39875912

Abstract

Background

Patients with obstructive sleep apnea (OSA) frequently experience sleep disturbance and psychological distress, such as depression and anxiety, which may have a negative impact on their health status and functional abilities. To gain a more comprehensive understanding of the symptoms of depression, anxiety, and sleep disturbance in patients with OSA, the current study utilized network analysis to examine the interconnections among these symptoms.

Methods

Depressive and anxiety symptoms were evaluated using the Hospital Anxiety and Depression Scale (HADS), and sleep disturbance symptoms were evaluated using the Pittsburgh Sleep Quality Index (PSQI). A total of 621 patients with OSA completed the questionnaires. The indices 'Expected influence' and 'Bridge expected influence' were used as centrality measures in the symptom network. The Least Absolute Shrinkage and Selection Operator (LASSO) technique and the Extended Bayesian Information Criterion (EBIC) were utilized to estimate the network structure of depressive, anxiety, and sleep disturbance symptoms. A Network Comparison Test (NCT) was performed to evaluate the differences between the mild to moderate OSA and severe OSA networks.

Results

Network analysis revealed that A6 (“Getting sudden feelings of panic”) had the highest expected influence value and D6 (“Feeling being slowed down”) had the highest bridge expected influence values in the networks. The NCT results revealed that the edge weights significantly differed between patients with mild to moderate OSA and those with severe OSA (M = 0.263, p = 0.008). There was no significant difference in global strength variation between the two networks (S = 0.185, p = 0.773).

Conclusions

Our results suggest that the highest expected influence value and bridge symptoms (e.g., A6 and D6) can be prioritized as potential targets for intervention and treatment in patients with OSA.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12888-025-06532-w.

Keywords: Obstructive sleep apnea, Depressive, Anxiety, Sleep disturbance, Network analysis

Introduction

Obstructive sleep apnea (OSA) is characterized by recurrent partial or complete collapse of the upper airway during sleep, resulting in either reduced (hypopnea) or absent (apnea) airflow for at least ten seconds [1]. This condition leads to significant changes in intrathoracic pressure, intermittent hypoxemia, and frequent sleep arousals. Although these arousals typically do not fully awaken the patient, they cause sleep fragmentation, reduced sleep efficiency, and consequently, daytime dysfunction, such as excessive daytime sleepiness [1, 2]. Research has demonstrated that chronic intermittent hypoxia associated with OSA may result in cognitive deficits, including impairments in intelligence, thinking abilities, and memory function [3]. Senaratna et al. [4] estimate that obstructive sleep apnea remains undiagnosed and untreated in approximately one billion adults worldwide, underscoring the substantial global burden of this condition.

Additionally, patients with OSA suffer from psychological distress, including depression and anxiety [5–8]. Lechat et al. [9] demonstrated that the chronic intermittent hypoxia experienced by OSA patients could increase the risk of neuroinflammation, which may contribute to the development of depressive and anxiety symptoms. According to a recent systematic review and meta-analysis study, the prevalence of depressive and anxiety symptoms in patients with OSA was 35% and 32%, respectively [6]. Previous research has suggested that daytime dysfunction is associated with depressive symptoms [10–12] and anxiety symptoms [12, 13], such as excessive daytime sleepiness, which may provide more accurate predictions of depression in patients with OSA [6], while depression and anxiety can contribute to daytime dysfunction [13, 14]. Clinical experience supports that the evaluation and treatment of psychological distress, as opposed to the disordered breathing itself, could significantly decrease a hallmark symptom of depression in patients with OSA [15]. Hence, a deeper understanding of the relationships among depression, anxiety, and sleep disturbance coupled with the appropriate application of assessment tools can substantially improve the precision of diagnosis and the efficacy of treatment for OSA patients.

The majority of research on the comorbidities of depression, anxiety, and sleep disturbances has relied primarily on syndromic assessments using standard scale total scores, which is a common approach to enhance understanding of patients' symptoms. This method is rooted in traditional psychopathological theories, which typically emphasize the association of individual symptoms with prototypical symptoms of other syndromes rather than highlighting the differences among individual symptoms themselves [16]. Notably, these scores are composed of various individual symptoms, each of which may involve different underlying psychoneurological mechanisms. Recently, network analysis has emerged as a novel approach for examining the complex and dynamic changing relationships among individual mental symptoms [17]. This approach is grounded in the notion that psychological distress consists of an intricate network of interrelated symptoms [18], with an emphasis on understanding the nature and intensity of the associations among these symptoms [19, 20]. Additionally, it has been demonstrated to be a valuable tool for understanding the interconnections among specific symptoms and diverse psychiatric disorders [21], provides a comprehensive and intuitive representation of the intricate relationships among the variables, and enables a deeper comprehension of the interplay between individual mental symptoms, which would be difficult to capture using total scores from standardized questionnaires [22]. Impressively, network analysis has the ability to capture and visually display the direct and indirect connections between variables from a data-driven standpoint [23]. Moreover, network analysis identifies the central and influential nodes within a network structure, providing valuable insights into the key variables that play a pivotal role in the overall network [19, 24].

Network analysis has been used to explore the relationships between the depressive, anxiety, and sleep disturbance symptoms. For instance, a network analysis was conducted to explore the relationship between sleep quality and social anxiety in breast cancer patients. The network structure revealed that “Subjective Sleep Quality” had the highest expected influence value within the network. Furthermore, “Working Difficulty Under Observation” and “Sleep Disorders” exhibited the highest bridge expected influence values within their respective communities [25]. Another study employed network analysis to investigate the network structure of symptoms associated with anxiety and sleep problems among outpatients in high-altitude areas. In this analysis, “Nervousness,” “Trouble Relaxing,” and “Uncontrollable Worry” were identified as the most critical central and bridge symptoms within the network structure of anxiety and sleep problems [26]. Network analysis was utilized to estimate the relationships among various aspects of sleep and mental health, revealing that depression, daytime dysfunction, and well-being acted as “bridges” connecting the domains of sleep and mental health in the network [27]. Few studies have utilized network analysis to investigate the correlation between depressive, anxiety, and sleep disturbance symptoms in patients with OSA. Conducting network analyses in patients with OSA could provide valuable insights into the complex relationships and potential causal pathways among these variables.

The present study presents a network analysis of the interrelationships among the depressive, anxiety, and sleep disturbance symptoms in patients with OSA. The aim of this study is to identify the core symptoms and bridge symptoms within the network and to assess the consistency of the two network structures measured in patients with mild to moderate OSA and severe OSA.

Methods

Participants and procedure

This study was conducted at the Sleep Medical Center, Nanfang Hospital, from October 18, 2022, to May 31, 2023. Patients with OSA who underwent polysomnography were invited to participate in this study after being assessed for eligibility by the attending psychiatrist. All participants were newly diagnosed patients who had not received any previous treatment for OSA. After agreeing to participate, patients were asked to sign an informed consent form and were instructed to answer all the questions in the questionnaire before submission. The data were automatically saved in the system upon submission.

The severity of OSA is commonly assessed using the apnea–hypopnea index (AHI), which is defined as the number of apneas and hypopneas per hour of sleep (or hour of recording for home tests). OSA was diagnosed when the AHI was more than 5 events per hour. Specifically, an AHI ranging from 5 to 14.9 events per hour indicates mild OSA, an AHI between 15 and 29.9 events per hour indicates moderate OSA, and an AHI of 30 or more events per hour indicates severe OSA [1]. This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Nanfang Hospital, Southern Medical University.

Measures

The survey collected demographic information and assessed of symptoms related to depression, anxiety, and sleep disturbance. The severity of depressive and anxiety symptoms was assessed using the Hospital Anxiety and Depression Scale (HADS), which is a 14-item questionnaire rated from “0” (not at all) to “3” (nearly every day) [28]. The Cronbach's alpha coefficients of the Chinese versions of the HADS-A and the HADS-D were 0.75 and 0.76, respectively, suggesting good internal consistency for assessing symptoms [29]. The Pittsburgh Sleep Quality Index (PSQI) is a self-administered questionnaire designed to assess sleep disturbances. The PSQI, comprising 19 individual items, generates scores across seven distinct dimensions: subjective sleep quality (SSQ), sleep latency (SL), sleep duration (SDu), habitual sleep efficiency (SE), sleep disturbances (SDis), Hypnotic drugs (HD), and daytime dysfunction (DD) [30]. Scoring for each dimension is based on a scale of 0 to 3, with the total score ranging from 0 to 21 points. Higher scores on this scale correspond to poorer sleep quality. The specific content for each scale and item is shown in Table S1 of the Supplementary Material.

Statistical analyses

Network estimation

The R program (version 4.3.0) was used to conduct network analyses, utilizing the bootnet and qgraph packages developed by S Epskamp, D Borsboom and EI Fried [31]. In this methodology, symptoms are represented as nodes, and the associations between symptoms are depicted as edges [19]. The network analysis included a total of 21 nodes, representing different symptoms. These nodes encompassed seven depressive, anxiety, and sleep disturbance symptoms for each category, as detailed in Table S1. Nodes in the network that are more concentrated and positioned closer to one another have stronger and more frequent interactions with other nodes.

To enhance the interpretability of the results and to mitigate the presence of spurious connections, the Least Absolute Shrinkage and Selection Operator (LASSO) technique was employed [31–33] to regularize the network models. When using the Extended Bayesian Information Criterion (EBIC) for model selection, the tuning parameter is typically set between 0 and 0.5 [34, 35].

To evaluate the significance of each node in the network, centrality indices, such as strength, betweenness, and closeness, were estimated using previously established methodologies [36–39]. These calculations were performed using the 'centralityPlot' function within the qgraph package [40]. The centrality values were normalized (z scored) after calculation to facilitate comparison across different nodes. The normalized centrality values were then visualized in a plot, allowing for a clear representation of the relative importance of each node within the network.

To identify specific symptoms that act as bridges connecting other symptoms in the network, the 'bridge' function from the networktools package was employed [33]. This function helps to identify nodes with high betweenness centrality, indicating their crucial role in mediating connections between other symptoms. Detecting these bridge symptoms will help identify the key nodes influencing the flow of information within the network [18, 41]. The network is plotted using the Fruchterman-Reingold algorithm [42]. The predictability of the network model nodes was estimated using the R package "mgm" [43].

Estimation of network accuracy and stability

To assess the resilience of the network solution, the accuracy of edge weights and the reliability of centrality indices were evaluated using the “bootnet” R package. To assess the accuracy of the edge weights, confidence intervals (CIs) were calculated using nonparametric bootstrapping [44, 45]. Following the initial analysis, the dataset was randomly resampled to generate new datasets, and 95% confidence intervals (CIs) were calculated from these resampled datasets. The width of the confidence intervals reflects the uncertainty associated with the estimated network parameters, with narrower CIs indicating a more reliable and stable network [31].

To compare variations in network characteristics, we conducted 1000 permutations and employed a bootstrap differential test. The discrepancies obtained from the bootstrap differential test were utilized to compare the edge weights and node centrality indices [46]. Subset bootstrapping involves randomly selecting subsets of the data and recalculating the centrality indices multiple times. By computing correlation stability coefficients (CS), we can quantify the stability of the centrality indices across these subsets [47]. When assessing the stability of centrality indices using the CS, it is generally recommended that they should not be lower than 0.25 and should preferably be above 0.50 [31].

Comparison of networks between mild to moderate OSA and severe OSA

Utilizing the R package “NetworkComparisonTest,” the NCT was performed to assess differences between mild to moderate OSA and severe OSA [48]. The NCT is a two-tailed permutation approach used to compare two networks by employing multiple invariance measures [48, 49].

The NCT was utilized to estimate differences between the two networks with respect to (1) global network strength, defined as the absolute sum of all edge weights and (2) network structure, which is characterized by the distribution of edge weights within each network and the differential strength of each edge when comparing the two networks, with adjustments made for multiple testing [38]. A p value less than 0.05 was considered to indicate statistical significance (two-tailed).

Results

A total of 621 patients with OSA were recruited in mainland China, of whom 308 were diagnosed with mild to moderate OSA and 313 were diagnosed with severe OSA. The percentage of males was 79.7%, and the mean age was 45.34 ± 13.65 years (mean ± SD). The mean scores for sleep disturbance and HADs were 9.69 ± 4.58 and 12.22 ± 7.89 (mean ± SD), respectively. Table 1 presents an overview of the demographic characteristics of the participants.

Table 1.

Demographic characteristics of patients with OSA

Variable Mean/N SD/%
Gender (male) 495 79.7
Age (years) 45.34 13.65
BMI (kg/m2) 26.06 3.83
Ethnicity (Han) 517 83.25
Marital status
 Married 93 14.9
 Unmarried 408 65.7
 Divorced 19 3
 Widowed 6 0.9
Occupation
 Student 16 2.5
 Medical personnel 15 2.4
 Civil servants / Government employees 36 5.7
 Company employees 128 20.6
 Service industry personnel 22 3.5
 Workers 24 3.8
 Agricultural, forestry, animal husbandry, fishery, and water conservancy production personnel 13 2
 Self-employed individuals 112 18
 Freelancers 75 12
 Retired individuals 57 9.1
 Unemployed individuals 28 4.5
OSA patient
 Mild OSA 156 25.1
 Moderate OSA 152 24.4
 Severe OSA 313 50.4
Mental health status
 Mild-Moderate OSA with HADS anxiety score ≥ 8 111 36.04
 Mild-Moderate OSA with HADS depression score ≥ 8 99 32.14
 Severe OSA with HADS anxiety score ≥ 8 74 23.64
 Severe OSA with HADS depression score ≥ 8 78 24.92
 PSQI 9.69 4.58
 HADS 12.22 7.89
 HADS_anxiety 5.99 4.51
 HADS_depression 5.87 4.16
Open in a new tab

Abbreviations: BMI body mass index, PSQI The Pittsburgh Sleep Quality Index, HADS Hospital anxiety and depression scale

Network structures

The analysis of the network consisting of 21 nodes, as shown in Fig. 1, revealed that a total of 120 out of 210 edges (57.14%) were estimated to be nonzero edges, indicating considerable interconnectedness between these symptoms. The three strongest edges were the between A2 and A6 (“Having a frightened feeling as if something awful is about to happen,”—“Getting sudden feelings of panic”), SDu and SE (“sleep duration,”—“habitual sleep efficiency”), and A3 and A6 (“Worrying thoughts go through mind”—“Getting sudden feelings of panic”). The strongest edges were distributed within the individual communities for depression, anxiety, and sleep disturbance, as shown in Fig. 1 and Supplementary Table S2.

Fig. 1.

Fig. 1

Open in a new tab

Symptom network of depression, anxiety and sleep quality symptoms in patients with obstructive sleep apnoea (OSA). In the diagram, symptom nodes with stronger connections are positioned closer to each other

The predictability values of the nodes ranged from 21.5% to 66.8%, with an average value of 43.9%, suggesting that approximately 43.9% of the variability in nodes within the network could be explained by their neighboring nodes (Table S1). A6 (“Getting sudden feelings of panic”) had the highest level of predictability. The SDis (“sleep disturbance”), symptoms of sleep disturbance, had the lowest level of predictability in the study (Table S1).

Figure 2 illustrates the centrality measures of all nodes in the study. The symptom with the highest strength centrality was D2 (“Unable to laugh and see the funny side of things”), followed by SSQ (“Subjective sleep quality”). Analyses of the expected influence centrality revealed that A6 (“Getting sudden feelings of panic”) had the most frequent occurrence on the shortest path between other nodes, followed by the SSQ (“Subjective sleep quality”) (as shown in Fig. 2). Vanek et al. [50] state that the panic and OSA have a bidirectional relationship, where each condition can exacerbate the other, necessitating a comprehensive treatment approach to improve both physical health and psychological well-being. Figure S5 illustrates the centrality indices within the networks of patients with mild to moderate OSA and those with severe OSA, indicating that A6 (“Getting sudden feelings of panic”) had the highest strength centrality in both networks.

Fig. 2.

Fig. 2

Open in a new tab

Centrality measures of all symptoms within the network. The illustration presents the centrality measures—strength, closeness, betweenness and expectedinfluence—of each symptom within the network (z-scores)

For bridge symptoms, D6 (“Feeling being slowed down”), DD (“Daytime dysfunction”), and D5 (“Not looking forward with enjoyment to things”) had the highest bridge strength values, and D6 (“Feeling being slowed down”), D5 (“Not looking forward with enjoyment to things”), and A1 (“Feeling tense or ‘wound up’”) also had high bridge expected influence values (Fig. S2).

Network accuracy and stability

Figure 3 illustrates the stability of the centrality indices using a case-dropping subset bootstrap analysis. The CS coefficients for strength and expected influence were 0.59 and 0.75, respectively (as shown in Fig. 3).

Fig. 3.

Fig. 3

Open in a new tab

Stability of centrality indices by case dropping subset bootstrap. The horizontal axis represents the proportion of the initial sample's cases incorporated at each step. The vertical axis captures the mean correlation coefficients between the centrality indices of the original network and those derived from networks following the progressive exclusion of case samples. The chromatic regions denote the 95% Confidence Intervals

To assess the accuracy and stability of the network estimation, a bootstrapping process consisting of 2500 iterations was employed, as shown in Figs. S1 and S3. Fig. S3 shows the results of the bootstrapped difference tests in the study. The edge weights of SDu-SE (“sitting and talking to someone”- “in a car”) and A2-A6 (“sitting in a public place”- “as a passenger in a car”) were significantly greater than those of the other edges.

The nonparametric bootstrapped difference test for strength demonstrated that node D2 (“Unable to laugh and see the funny side of things”) had the greatest strength compared to the other nodes (Fig. S4).

Network comparisons

Figure 4 displays the results of the network comparison tests. The results revealed that there was no significant difference in global strength between mild to moderate OSA patients and severe OSA patients (S = 0.185, p = 0.773), but there was a significant difference in the edge weights (M = 0.263, p = 0.008) (as shown in Fig. 4). The coefficient of similarity was S = 0.185, suggesting strong similarity between the two networks.

Fig. 4.

Fig. 4

Open in a new tab

Network comparisons test results. A Network edge invariance test. B Network difference in global strength

The Bonferroni-Holm correction was used to assess the potential different edges between the two networks. The edges “SE-SDis” (“Habitual sleep efficiency”- “Sleep disturbance”), “A7-D5” (“Having frightened feeling like ‘butterflies’ in the stomach”-“Not looking forward with enjoyment to things”), “DD-A6” (“Daytime dysfunction”-“Getting sudden feelings of panic”), “SSQ-SE” (“Subjective sleep quality”-“Habitual sleep efficiency”), and “SL-DD” (“Sleep latency”-“Daytime dysfunction”) showed statistically significant differences (p < 0.001) (as shown in Fig. 5). Mild to moderate OSA did not present edges of “SE-SDis” (with a strength of 0.179 in the severe OSA network) or “A7-D5” (with a strength of 0.147 in the severe OSA network), while severe OSA did not have the edges “DD-A6” (with a strength of 0.056 in the mild to moderate OSA network), “SSQ-SE”(with a strength of 0.277 in the mild to moderate OSA network), or “SL-DD” (with a strength of 0.184 in the mild to moderate OSA network).

Fig. 5.

Fig. 5

Open in a new tab

The depression, anxiety and sleep quality network model in OSA patients. A. mild to moderate OSA network; (B). severe OSA network. In the diagram, symptom nodes with stronger connections are positioned closer to each other

Discussion

To the best of our knowledge, this study is the first to utilize network analysis to investigate the correlation between symptoms of depression, anxiety, and sleep disturbance in patients with OSA. The obtained network model revealed notable stability and demonstrated associations among depression, anxiety, and sleep disturbance. The results indicate that there were no significant differences in overall network connectivity between the networks of patients with mild to moderate OSA and those with severe OSA.

Our analysis revealed that the strongest edges resided within a specific disorder community rather than between different disorder communities. This finding aligns with the results reported in previous network analysis studies, reinforcing the consistency of these observations across different investigations [19, 51, 52]. The present study revealed that the most direct connection was between anxiety symptoms A2 (“Having frightened feeling as if something awful is about to happen”) and A6 (“Getting sudden feelings of panic”). The second connection was between the SDu and SE (“sleep duration”—“habitual sleep efficiency”); the third connection was between A3 and A6 (“Worrying thoughts go through mind”—“Getting sudden feelings of panic”). A recent network analysis of anxiety-depressive symptoms in Chinese rheumatoid arthritis patients showed that the strongest edge within the anxiety symptoms was between A2 (“Afraid something will happen”) and A6 (“Sudden panic”) [53]. The replication of this strong connection across different patient populations suggests that the link between these two symptoms may reflect a more general feature of anxiety symptom networks, rather than being specific to OSA. The presence of OSA may contribute to the strength of this connection. OSA is characterized by repeated episodes of complete or partial upper airway obstruction during sleep, leading to physiological responses such as breath-holding, feelings of suffocation, and rapid heart rate. These experiences may heighten the association between frightening feelings and sudden panic in OSA patients, as the physiological arousal caused by apneic events could serve as a trigger for both symptoms.

Nodes with high expected influence centrality are effective at identifying symptoms that contribute most to the overall symptom network. The anxiety symptom A6 (“getting sudden feelings of panic”) exhibited the highest centrality value in the network. This indicated that node A6 had the most significant influence on the entire symptom network. Consequently, an intervention aimed at attenuating sudden feelings of panic may also alleviate other symptoms in the depressive-anxiety-sleep disturbance symptom network. The primary recommended treatment for obstructive sleep apnea (OSA) is continuous positive airway pressure (CPAP) therapy [54]. CPAP is a highly effective intervention that provides a steady stream of air through a mask to keep the airway open during sleep, thus preventing the breathing pauses characteristic of OSA. Long-term CPAP use has been shown to improve sleep quality, reduce daytime sleepiness, and lower the risk of cardiovascular complications in OSA patients [55–57]. However, adherence to CPAP therapy can be challenging for some patients, particularly those with comorbid psychological symptoms such as anxiety and panic. To address this issue and enhance CPAP adherence, cognitive-behavioral therapy (CBT) is recommended as an adjunctive treatment for OSA patients. Studies have shown CBT to be effective in reducing panic attacks and improving quality of life, with benefits maintained over the long-term [58, 59]. CBT can help patients learn skills to manage panic symptoms and negative emotions, change maladaptive thinking patterns and behaviors, and gradually re-establish normal functioning [60, 61]. Moreover, Pataka et al. [62] and Sforza et al. [63] have shown that CBT significantly improves adherence to CPAP therapy, suggesting that addressing psychological symptoms, particularly panic, can enhance overall treatment success in OSA patients. By targeting the high centrality symptom of sudden feelings of panic with CBT alongside CPAP therapy, there is potential to positively influence the entire symptom network and improve both psychological well-being and adherence to the primary OSA treatment.

The presence of bridging symptoms between different disorder communities significantly influences comorbidities. It is essential to identify bridge symptoms that play a pivotal role in mitigating contagion between psychiatric syndromes within network models and to prioritize these symptoms as targets for treatment [18]. Node D6 (“Feeling being slowed down”), D5 (“Not looking forward with enjoyment to things”) and A1 (“Feeling tense or ‘wound up’”) was identified as a bridge symptom in the depressive-anxiety-sleep disturbance symptom network. A hierarchical therapeutic approach targeting these bridge symptoms may effectively disrupt the interconnected network of depression, anxiety, and sleep disturbance. CBT serves as the first-line psychological intervention, particularly effective for addressing psychomotor retardation (D6) through behavioral activation techniques and cognitive restructuring [64]. For anhedonia symptoms (D5), Behavioral Activation therapy, a key component of CBT, systematically helps patients re-engage in pleasurable activities and rebuild reward sensitivity [64]. Tension and arousal symptoms (A1) are primarily addressed through relaxation training, including progressive muscle relaxation and mindfulness-based stress reduction techniques [65]. When psychological interventions alone prove insufficient, selective serotonin reuptake inhibitors (SSRIs) may be considered as second-line treatment, particularly effective for both psychomotor retardation and anhedonia [66]. Additionally, patients are also advised to lead a healthy lifestyle in their daily lives, which includes a good diet, adequate sleep, and moderate exercise [67, 68].

This study identified A6 (“getting sudden feelings of panic”) and D6 (“Feeling being slowed down”) as the central and key bridge symptom in the depressive-anxiety-sleep disturbance network in patients with OSA in China. This study suggested that a hierarchical therapeutic approach combining psychological interventions, pharmacotherapy, and lifestyle modifications may effectively target these symptoms and may yield beneficial outcomes in ameliorating concurrent manifestations of depression, anxiety, and sleep disturbance among patients with OSA.

Strength and limitations

This study has several strengths. First, the patients with OSA were strictly screened, which guaranteed the reliability of the results. Second, network analysis of these symptoms can help identify nodes of importance in the disease network, e.g., symptoms with high connectivity or common biomarkers. This can help to identify potential therapeutic targets and key predictors of disease progression.

However, several limitations need to be addressed. First, the findings were based on self-reported scales, which may have been influenced by recall bias and social desirability bias. Additionally, since our study was cross-sectional, causal relationships cannot be established between the symptoms. This study did not include a non-OSA control group, which makes it difficult to determine if the observed network characteristics are specific to OSA or reflect more general properties of psychological symptom networks. Future studies might consider comparing network structures between OSA patients and healthy controls to identify OSA-specific symptom patterns.

Conclusions

In summary, our study employed network analysis to reveal the complex relationships among depression, anxiety, and sleep disturbance in patients with OSA. These findings contribute to a deeper understanding of psychological issues in OSA patients and provide guidance for the development of personalized treatment strategies. Future research could further explore the causal relationships among these psychological symptoms and expand the sample size to enhance the reliability and generalizability of the study results.

Supplementary Information

Supplementary Material 1. (1,012.7KB, docx)

Acknowledgements

We would like to thank all participants in this study.

Abbreviations

OSA

Obstructive sleep apnea

HADS

The Hospital Anxiety and Depression Scale

PSQI

The Pittsburgh Sleep Quality Index

LASSO

The Least Absolute Shrinkage and Selection Operator

EBIC

The Extended Bayesian Information Criterion

NCT

Network Comparison Test

AHI

The apnea–hypopnea index

CIs

Confidence intervals

CS

Correlation stability coefficients

SD

Standard deviation

CBT

Cognitive behavioral therapy

SSRIs

5-Hydroxytryptamine reuptake inhibitors

Authors’ contributions

XL: Conceptualization, Methodology, Validation, Formal analysis, Writing–original draft, Writing–review & editing. SL: Conceptualization. QW: Formal analysis. YX: Data curation, Formal analysis. RF: Investigation, Validation. YC: Investigation, Validation. BZ: Conceptualization, Writing– review & editing, Supervision, Project administration, Funding acquisition. All authors reviewed the manuscript.

Funding

This work was supported by the National Key R&D Program of China [grant number 2021YFC2501500], National Natural Science Foundation of China [grant number 82271525, 82071488], Nanfang Hospital Clinical Research Project of Southern Medical University [grant number 2021CR009], and Natural Science Foundation of Guangdong Province [grant number 2023A1515010456].

Data availability

Data and materials are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Nanfang Hospital, Southern Medical University. All participants provided informed consent after the study procedures were described to them.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Xue Luo and Shuangyan Li contributed equally to this work.

References

  • 1.Gottlieb DJ, Punjabi NM. Diagnosis and management of obstructive sleep apnea: a review. JAMA. 2020;323(14):1389–400. [DOI] [PubMed] [Google Scholar]
  • 2.Steier JS, Bogan RK, Cano-Pumarega IM, Fleetham JA, Insalaco G, Lal C, Pépin JL, Randerath WJ, Redline S, Malhotra A. Recommendations for clinical management of excessive daytime sleepiness in obstructive sleep apnoea - A Delphi consensus study. Sleep Med. 2023;112:104–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Gupta A, Tripathi A, Sharma P. The long-term effects of mandibular advancement splint on cardiovascular fitness and psychomotor performance in patients with mild to moderate obstructive sleep apnea: a prospective study. Sleep Breath. 2017;21(3):781–9. [DOI] [PubMed] [Google Scholar]
  • 4.Senaratna CV, Lowe A, Walters EH, Abramson MJ, Bui D, Lodge C, Erbas B, Burgess J, Perret JL, Hamilton GS, et al. Associations of early life and childhood risk factors with obstructive sleep apnoea in middle-age. Respirology. 2024;29(1):63–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Vanek J, Prasko J, Genzor S, Ociskova M, Kantor K, Holubova M, Slepecky M, Nesnidal V, Kolek A, Sova M. Obstructive sleep apnea, depression and cognitive impairment. Sleep Med. 2020;72:50–8. [DOI] [PubMed] [Google Scholar]
  • 6.Garbarino S, Bardwell WA, Guglielmi O, Chiorri C, Bonanni E, Magnavita N. Association of anxiety and depression in obstructive sleep apnea patients: a systematic review and meta-analysis. Behav Sleep Med. 2020;18(1):35–57. [DOI] [PubMed] [Google Scholar]
  • 7.Gharsalli H, Harizi C, Zaouche R, Sahnoun I, Saffar F, Maalej S, El Gharbi LD. Prevalence of depression and anxiety in obstructive sleep apnea. Tunis Med. 2022;100(7):525–33. [PMC free article] [PubMed] [Google Scholar]
  • 8.Duan X, Zheng M, Zhao W, Huang J, Lao L, Li H, Lu J, Chen W, Liu X, Deng H. Associations of depression, anxiety, and life events with the risk of obstructive sleep apnea evaluated by berlin questionnaire. Front Med (Lausanne). 2022;9:799792. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Lechat B, Scott H, Manners J, Adams R, Proctor S, Mukherjee S, Catcheside P, Eckert DJ, Vakulin A, Reynolds AC. Multi-night measurement for diagnosis and simplified monitoring of obstructive sleep apnoea. Sleep Med Rev. 2023;72:101843. [DOI] [PubMed] [Google Scholar]
  • 10.Björnsdóttir E, Benediktsdóttir B, Pack AI, Arnardottir ES, Kuna ST, Gíslason T, Keenan BT, Maislin G, Sigurdsson JF. The prevalence of depression among untreated obstructive sleep apnea patients using a standardized psychiatric interview. J Clin Sleep Med. 2016;12(1):105–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.LaGrotte C, Fernandez-Mendoza J, Calhoun SL, Liao D, Bixler EO, Vgontzas AN. The relative association of obstructive sleep apnea, obesity and excessive daytime sleepiness with incident depression: a longitudinal, population-based study. Int J Obes (Lond). 2016;40(9):1397–404. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Lee SA, Han SH, Ryu HU. Anxiety and its relationship to quality of life independent of depression in patients with obstructive sleep apnea. J Psychosom Res. 2015;79(1):32–6. [DOI] [PubMed] [Google Scholar]
  • 13.Hasler G, Buysse DJ, Gamma A, Ajdacic V, Eich D, Rössler W, Angst J. Excessive daytime sleepiness in young adults: a 20-year prospective community study. J Clin Psychiatry. 2005;66(4):521–9. [DOI] [PubMed] [Google Scholar]
  • 14.Lin CY, Imani V, Griffiths MD, Broström A, Nygårdh A, Demetrovics Z, Pakpour AH. Temporal associations between morningness/eveningness, problematic social media use, psychological distress and daytime sleepiness: Mediated roles of sleep quality and insomnia among young adults. J Sleep Res. 2021;30(1):e13076. [DOI] [PubMed] [Google Scholar]
  • 15.Bardwell WA, Moore P, Ancoli-Israel S, Dimsdale JE. Fatigue in obstructive sleep apnea: driven by depressive symptoms instead of apnea severity? Am J Psychiatry. 2003;160(2):350–5. [DOI] [PubMed] [Google Scholar]
  • 16.Borsboom D. Psychometric perspectives on diagnostic systems. J Clin Psychol. 2008;64(9):1089–108. [DOI] [PubMed] [Google Scholar]
  • 17.Fisher AJ, Reeves JW, Lawyer G, Medaglia JD, Rubel JA. Exploring the idiographic dynamics of mood and anxiety via network analysis. J Abnorm Psychol. 2017;126(8):1044. [DOI] [PubMed] [Google Scholar]
  • 18.Jones PJ, Ma R, McNally RJ. Bridge centrality: a network approach to understanding comorbidity. Multivar Behav Res. 2021;56(2):353–67. [DOI] [PubMed] [Google Scholar]
  • 19.Beard C, Millner AJ, Forgeard MJ, Fried EI, Hsu KJ, Treadway MT, Leonard CV, Kertz S, Björgvinsson TJPM. Network analysis of depression and anxiety symptom relationships in a psychiatric sample. Psychol Med. 2016;46(16):3359–69. [DOI] [PMC free article] [PubMed]
  • 20.Rouquette A, Pingault J-B, Fried EI, Orri M, Falissard B, Kossakowski JJ, Vitaro F, Tremblay R, Cote SM, Borsboom D. Emotional and behavioral symptom network structure in elementary school girls and association with anxiety disorders and depression in adolescence and early adulthood: a network analysis. JAMA PSYCHIAT. 2018;75(11):1173–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Borsboom D, Cramer AO. Network analysis: an integrative approach to the structure of psychopathology. Ann Rev Clin Psych. 2013;9:91–121. [DOI] [PubMed] [Google Scholar]
  • 22.Sardella A, Varallo G, Agus M, Lenzo V, Zagaria A, Terrone G, Musetti A, Castelnuovo G, Quattropani MC, Franceschini C. Anxiety, depression, emotion regulation, and daytime sleepiness: are there links between these factors? network analysis on an italian sample during the covid-19 pandemic. Clin Neuropsychiatry. 2023;20(5):395–404. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Epskamp S, Fried EIJPM. A tutorial on regularized partial correlation networks. Psychol Methods. 2018;23(4):617. [DOI] [PubMed]
  • 24.Contreras A, Nieto I, Valiente C, Espinosa R, Vazquez C. The study of psychopathology from the network analysis perspective: a systematic review. Psychother Psychosom. 2019;88(2):71–83. [DOI] [PubMed] [Google Scholar]
  • 25.He C, He Y, Yang T, Wu C, Lin Y, Yan J, Chang W, Chang F, Wang Y, Wu S. Relationship of sleep-quality and social-anxiety in patients with breast cancer: a network analysis. BMC Psychiatry. 2023;23(1):887. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Jin Y, Li J, Ye J, Luo X, Wilson A, Mu L, Zhou P, Lv Y, Wang Y. Mapping associations between anxiety and sleep problems among outpatients in high-altitude areas: a network analysis. BMC Psychiatry. 2023;23(1):341. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Yu B, Fu Y, Dong S, Reinhardt JD, Jia P, Yang S. Identifying potential action points for improving sleep and mental health among employees: a network analysis. Sleep Med. 2024;113:76–83. [DOI] [PubMed] [Google Scholar]
  • 28.Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the Hospital Anxiety and Depression Scale. An updated literature review. J Psychosom Res. 2002;52(2):69–77. [DOI] [PubMed] [Google Scholar]
  • 29.Yang Y, Ding R, Hu D, Zhang F, Sheng L. Reliability and validity of a Chinese version of the HADS for screening depression and anxiety in psycho-cardiological outpatients. Compr Psychiatry. 2014;55(1):215–20. [DOI] [PubMed] [Google Scholar]
  • 30.Buysse DJ, Reynolds CF III, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res. 1989;28(2):193–213. [DOI] [PubMed] [Google Scholar]
  • 31.Epskamp S, Borsboom D, Fried EI. Estimating psychological networks and their accuracy: a tutorial paper. Behav Res Methods. 2018;50(1):195–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Friedman J, Hastie T, Tibshirani R. Sparse inverse covariance estimation with the graphical lasso. Biostatistics. 2008;9(3):432–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Heeren A, Jones PJ, McNally RJ. Mapping network connectivity among symptoms of social anxiety and comorbid depression in people with social anxiety disorder. J Affect Disord. 2018;228:75–82. [DOI] [PubMed] [Google Scholar]
  • 34.Epskamp SJAPA. Regularized Gaussian psychological networks: Brief report on the performance of extended BIC model selection. 2016.
  • 35.van Borkulo C, Boschloo L, Borsboom D, Penninx BW, Waldorp LJ, Schoevers RA. Association of symptom network structure with the course of depression. JAMA Psychiat. 2015;72(12):1219–26. [DOI] [PubMed] [Google Scholar]
  • 36.Borgatti SP. Centrality and network flow. Soc Netw. 2005;27(1):55–71. [Google Scholar]
  • 37.Freeman LC. Centrality in social networks: Conceptual clarification. Soc Netw. 2002;1(3):215–39. [Google Scholar]
  • 38.Opsahl T, Agneessens F, Skvoretz J. Node centrality in weighted networks: generalizing degree and shortest paths. Soc Netw. 2010;32(3):245–51. [Google Scholar]
  • 39.Pan H, Liu Q. Difference of depression between widowed and non-widowed older people in China: a network analysis approach. J Affect Disord. 2021;280(Pt B):68–76. [DOI] [PubMed] [Google Scholar]
  • 40.Epskamp S, Cramer AO, Waldorp LJ, Schmittmann VD, Borsboom D. qgraph: network visualizations of relationships in psychometric data. J Stat Softw. 2012;48:1–18. [Google Scholar]
  • 41.Wang Y, Hu Z, Feng Y, Wilson A, Chen R. Changes in network centrality of psychopathology symptoms between the COVID-19 outbreak and after peak. Mol Psychiatry. 2020;25(12):3140–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Epskamp S, Waldorp LJ, Mõttus R, Borsboom D. The Gaussian graphical model in cross-sectional and time-series data. Multivar Behav Res. 2018;53(4):453–80. [DOI] [PubMed] [Google Scholar]
  • 43.Haslbeck J, Waldorp LJ: mgm: Estimating time-varying mixed graphical models in high-dimensional data. arXiv preprint arXiv. 2015:151006871.
  • 44.Chernick MR. Bootstrap methods: a guide for practitioners and researchers. Hoboken, NJ: John Wiley & Sons; 2011.
  • 45.Cheung JM, Jarrin DC, Ballot O, Bharwani AA, Morin CM. A systematic review of cognitive behavioral therapy for insomnia implemented in primary care and community settings. Sleep Med Rev. 2019;44:23–36. [DOI] [PubMed] [Google Scholar]
  • 46.Mullarkey MC, Marchetti I, Beevers CG. Using network analysis to identify central symptoms of adolescent depression. J Clin Child Adolesc Psychol. 2019;48(4):656–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Costenbader E, Valente TW. The stability of centrality measures when networks are sampled. Soc Netw. 2003;25(4):283–307. [Google Scholar]
  • 48.Van Borkulo CD, van Bork R, Boschloo L, Kossakowski JJ, Tio P, Schoevers RA, Borsboom D, Waldorp LJ. Comparing network structures on three aspects: a permutation test. Psychol Methods. 2023;28(6):1273–85. [DOI] [PubMed]
  • 49.Dalege J, Borsboom D, van Harreveld F, van der Maas HLJSP, science p. Network analysis on attitudes: a brief tutorial. Soc Psychol Personal Sci. 2017;8(5):528–37. [DOI] [PMC free article] [PubMed]
  • 50.Vanek J, Belohradova Minarikova K, Prasko J, Genzor S, Subova D, Hodny F, Pobeha P, Bocek J, Ociskova M. Obstruction sleep apnoea and panic disorder. Neuro Endocrinol Lett. 2023;44(2):86–96. [PubMed] [Google Scholar]
  • 51.Garabiles MR, Lao CK, Xiong Y, Hall BJ. Exploring comorbidity between anxiety and depression among migrant Filipino domestic workers: a network approach. J Affect Disord. 2019;250:85–93. [DOI] [PubMed] [Google Scholar]
  • 52.Groen RN, Ryan O, Wigman JT, Riese H, Penninx BW, Giltay EJ, Wichers M, Hartman CA. Comorbidity between depression and anxiety: assessing the role of bridge mental states in dynamic psychological networks. BMC Med. 2020;18:1–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Zhang L, Zhu W, Wu B. Network analysis of depression and anxiety symptoms in Chinese rheumatoid arthritis patients. PeerJ. 2023;11:e16356. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Bonsignore MR, Lombardi C, Lombardo S, Fanfulla F. Epidemiology, physiology and clinical approach to sleepiness at the wheel in osa patients: a narrative review. J Clin Med. 2022;11(13):3691. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Crawford-Achour E, Dauphinot V, Martin MS, Tardy M, Gonthier R, Barthelemy JC, Roche F. Protective effect of long-term cpap therapy on cognitive performance in elderly patients with severe OSA: the PROOF study. J Clin Sleep Med. 2015;11(5):519–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Batool-Anwar S, Goodwin JL, Kushida CA, Walsh JA, Simon RD, Nichols DA, Quan SF. Impact of continuous positive airway pressure (CPAP) on quality of life in patients with obstructive sleep apnea (OSA). J Sleep Res. 2016;25(6):731–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Fava C, Dorigoni S, Dalle Vedove F, Danese E, Montagnana M, Guidi GC, Narkiewicz K, Minuz P. Effect of CPAP on blood pressure in patients with OSA/hypopnea a systematic review and meta-analysis. Chest. 2014;145(4):762–71. [DOI] [PubMed] [Google Scholar]
  • 58.Pompoli A, Furukawa TA, Efthimiou O, Imai H, Tajika A, Salanti G. Dismantling cognitive-behaviour therapy for panic disorder: a systematic review and component network meta-analysis. Psychol Med. 2018;48(12):1945–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Strauss AY, Kivity Y, Huppert JD. Emotion regulation strategies in cognitive behavioral therapy for panic disorder. Behav Ther. 2019;50(3):659–71. [DOI] [PubMed] [Google Scholar]
  • 60.Leichsenring F, Heim N, Steinert C. A review of anxiety disorders. JAMA. 2023;329(15):1315–6. [DOI] [PubMed] [Google Scholar]
  • 61.Magnon V, Dutheil F, Vallet GT. Benefits from one session of deep and slow breathing on vagal tone and anxiety in young and older adults. Sci Rep. 2021;11(1):19267. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Pataka A, Kotoulas SC, Gavrilis PR, Karkala A, Tzinas A, Stefanidou A. Adherence to CPAP treatment: can mindfulness play a role? Life (Basel, Switzerland). 2023;13(2):296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Sforza M, Salibba A, Carollo G, Scarpellino A, Bertone JM, Zucconi M, Casoni F, Castronovo V, Galbiati A, Ferini-Strambi L. Boosting obstructive sleep apnea therapy by non-pharmacological approaches: a network meta-analysis. Sleep Med. 2024;115:235–45. [DOI] [PubMed] [Google Scholar]
  • 64.Dimidjian S, Barrera M Jr, Martell C, Muñoz RF, Lewinsohn PM. The origins and current status of behavioral activation treatments for depression. Annu Rev Clin Psychol. 2011;7:1–38. [DOI] [PubMed] [Google Scholar]
  • 65.Muhammad Khir S, Wan MohdYunus WMA, Mahmud N, Wang R, Panatik SA, MohdSukor MS, Nordin NA. Efficacy of progressive muscle relaxation in adults for stress, anxiety, and depression: a systematic review. Psychol Res Behav Manag. 2024;17:345–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.MacGillivray S, Arroll B, Hatcher S, Ogston S, Reid I, Sullivan F, Williams B, Crombie I. Efficacy and tolerability of selective serotonin reuptake inhibitors compared with tricyclic antidepressants in depression treated in primary care: systematic review and meta-analysis. BMJ. 2003;326(7397):1014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Zhao JL, Jiang WT, Wang X, Cai ZD, Liu ZH, Liu GR. Exercise, brain plasticity, and depression. CNS Neurosci Ther. 2020;26(9):885–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Gujral S, Aizenstein H, Reynolds CF 3rd, Butters MA, Erickson KI. Exercise effects on depression: possible neural mechanisms. Gen Hosp Psychiatry. 2017;49:2–10. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1. (1,012.7KB, docx)

Data Availability Statement

Data and materials are available from the corresponding author upon reasonable request.

Articles from BMC Psychiatry are provided here courtesy of BMC

RESOURCES