Depression and anxiety symptom network structure among patients with coronary heart disease and association with quality of life: protocol for a multicentre cross-sectional and prospective longitudinal study

Abstract

Background

Anxiety and depression are critical mental health problems among persons with coronary heart disease (CHD). The range of symptoms is an important stressor for adverse cardiovascular events, and these symptoms can be involved in various ways during the course of CHD. However, the characteristics and mechanisms of comorbidity between the two mental states from the viewpoint of symptom interactions in patients with CHD remain unclear. Therefore, we aim to apply a symptom-oriented approach to identify core and bridge symptoms between anxiety and depression in a population with CHD and to identify differences in network structure over time and symptomatic link profiles.

Methods and analysis

We designed a multicentre, cross-sectional, longitudinal study of anxiety and depression symptoms among patients with CHD. We will evaluate degrees of symptoms using the Generalized Anxiety Disorder Scale, the Patient Health Questionnaire and the WHO Quality of Life-Brief version. Patients will be followed up for 1, 3 and 6 months after baseline measurements. We will analyse and interpret network structures using R software and its packages. The primary outcomes of interest will include centrality, bridge connections, estimates, differences in network structures and profiles of changes over time. The secondary outcome measures will be the stability and accuracy of the network. By combining cross-sectional and longitudinal analyses, this study should elucidate the central and potential causative pathways among anxiety and depression symptom networks as well as their temporal stability in patients with CHD.

Ethics and dissemination

The project conforms to the ethical principles enshrined in the Declaration of Helsinki (2013 amendment) and all local ethical guidelines. The ethics committee at the University of South China approved the study (Approval ID: 2023-USC-HL-414). The findings will be published and presented at conferences for widespread dissemination.

Trial registration number

ChiCTR2300075813.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Data evaluated at four timepoints facilitated the understanding of the dynamic causal relationship between anxiety and depression symptoms in patients with coronary heart disease (CHD).

• The inter-relatedness of quality of life and CHD will be analysed.

• The nature of the study design includes possible biases such as loss to follow-up and other uncontrollable factors.

• The universal scale used to facilitate measurements might not fully identify individualised symptoms of anxiety and depression in patients with CHD.

• Because the samples will be collected at local hospitals, our findings might not be generalisable to other regions or populations.

Introduction

Widespread coronary heart disease (CHD) has emerged as a major cause of the global increase in the prevalence and mortality of cardiovascular diseases (CVD).^{1 2} Epidemiological findings of CVD suggest that the incidences of CHD in persons aged <45 and >45 years are increasing and decreasing, respectively.³ Recent evidence suggests that recurrent adverse CHD cardiovascular events account for 77% of secondary patient admissions within 30 days.^{4 5} In addition to somatic problems, patients with CHD also experience a series of psychological disorders. Depression and anxiety are prominent and prevalent mental symptoms among patients with CHD.^{6 7} The findings of a systematic review indicate that the global proportions of patients with cardiac disease who have depression and anxiety are 31.3% and 32.9%. Such high prevalence rates warrant more professional methods to control these disorders.⁸ Meta-analyses have linked depression and anxiety to decreased medication adherence, poor health behaviours and adverse cardiac events such as angina and acute myocardial infarction.^{9 10} Depression is an independent risk factor for an adverse cardiac prognosis. Specifically, anxiety often occurs in conjunction with depression.¹¹ Furthermore, susceptibility to developing depression is higher among patients with CHD and symptoms of anxiety than those without symptoms of anxiety.¹² The fact that depression and anxiety can mutually predict each other over periods from weeks to years has been established.¹³ Moreover, in terms of psychopathology, the comorbidity of mental symptoms being more likely a rule rather than an exception is obvious.¹¹ This could result in depression and anxiety forming a complicated coinfluencing relationship that affects the rehabilitation and quality of life (QOL) of patients with CHD.^{11 14} The focus of investigations has gradually transitioned from models of managing a single CHD symptom to multidimensional symptom clusters in patients. However, reports on the microstructure and mechanical indicators within the symptom, such as density, intensity, proximity and centrality, remain scant.^{15 16} In order to comprehensively and accurately reflect symptom occurrence profiles of patients with CHD, advanced methods are needed to determine interplay between pairs of symptoms, as well as between symptomatic clusters.¹⁷ Further, the results of clinical intervention studies have shown that the inter-relationships of several symptoms in patients with CHD lead to a lack of pertinence and precision in clinical intervention.^18–20

The symptom network is an advanced method of analysis, evolving from sociology, that has been applied to explore and visualise dynamic causal relationships between symptoms in psychopathology.²¹ Unlike the traditional description of mental disorders and symptoms, the symptom network model considers mental disorders as a part of a causal system involving symptom interactions rather than an underlying factor that independently causes symptoms.^{21 22} The central viewpoint is that symptoms constitute mental disorders, rather than merely reflecting them.²³ The symptomatic concept introduced in 2017²⁴ suggested that clusters of symptoms might present various network characteristics. For example, in the network of depression, ‘anhedonia’ might be closely related to ‘sad mood’²⁵ or ‘hopelessness’.²⁶ This concept redefines the causal relationship between mental disorders and symptoms and provides a new perspective for exploring intersymptom profiles and explaining psychopathological phenomena. In the classical symptom network, nodes correspond to symptoms, and edges represent associations between symptoms.²⁷ Furthermore, the key role of symptoms can be expressed through characteristic indicators. Hysteresis, a distinct hallmark of the network model, explains that the symptom network can persist even if the triggering factors disappear under certain conditions.^{24 27 28} This feature might be well adapted to the network of anxiety and depression because they are closely interconnected and various characteristics can manifest at different stages of the disease process.^28–30 Therefore, a network composed of longitudinal data has the potential to deepen understanding of changes in profiles of patients with anxiety and depression, particularly through the medium of hysteresis.³¹ We designed a multicentre, cross-sectional and prospective longitudinal study to develop a symptom network model for anxiety and depression in a CHD population. In each wave survey, we plan to identify the core symptoms and unknown associations among symptom clusters of anxiety and depression in patients with CHD. We will attempt to provide information that is difficult to explain solely based on the incidence and severity of symptoms. Recognising the limitations of cross-sectional data, we will conduct longitudinal studies to discern temporal prediction and directionality and analyse the mechanism of action between influencing factors, symptoms and outcomes. Our goal is to identify and measure the central symptoms, symptom network structure and bridge connections between anxiety and depression in patients with CHD that can help to explain the causal pathways of specific symptoms. The revealed profiles might facilitate a more comprehensive understanding of the relationship between anxiety and depression in such patients and provide meaningful references for future clinical care.

Methods and analysis

Study design

This multicentre cross-sectional and observational longitudinal study will aim to identify the central and outstanding symptoms of depression and anxiety in patients with CHD. This study intends to serve as a reference for developing targeted interventions to improve the quality of nursing care provided by clinical nurses. A longitudinal meta-analysis found that the impact of anxiety on depressive symptoms and vice versa diminishes over time.¹³ Thus, we will use symptom measures effects to examine shorter durations, enabling the assessment of how moods can influence each other and elicit significant interaction effects within short periods. Anxiety and depression symptoms tend to stabilise over time, and interactions between them in terms of predicting recurrent cardiac events are moderated by symptoms of depression, and these might manifest in patients with acute cardiac diseases within 6 months.^{32 33} The data for primary outcomes at baseline (T0) will be collected between November 2023 and February 2024, patients will then be assessed at 1 (T1), 3 (T2) and 6 (T3) months of follow-up. Figure 1 shows the flow of patients through the study.

Figure 1.

Flow chart of this study.

Sample size

The sample size of the symptom network depends on the number of nodes in the network structure, since the parameters to be estimated are influenced by changes in the nodes. The number of participants should ideally be equal to the number of parameters in the network.²⁷ Using the formula proposed in that report and considering the number of variables included in our study design, the estimated network should have a maximum of 16 nodes and 136 parameters (16×15)/2+16, according to having at least 3‒5 patients per parameter. Thus, the sample should contain at least 408 participants. Accounting for a 20% loss during follow-up, the sample size will be expanded by 20% of the original size. Therefore, a total of 490 participants will be required. All institutions participating in this study are tertiary teaching hospitals with an annual inpatient capacity for cardiovascular medicine of 4800‒17 000. Considering the feasibility of implementation, we will try to ensure that the number of assessments can sustain ~490 participants at all measurement times, accounting for the sample attrition rate.

Setting and recruitment

This study will be conducted at five tertiary hospitals in four cities: three of these hospitals are affiliated with the University of South China in Hengyang and Changsha Hunan Province and one each is associated with Guangdong Provincial People’s Hospital in Guangzhou and ShunDe Hospital Guangzhou University of Chinese Medicine in Foshan, both in Guangdong province. All of these institutions are major public tertiary hospitals with comprehensive management systems, ensuring that every inpatient is included in the electronic medical record system. Patients hospitalised in the cardiovascular medicine departments between November 2023 and February 2024, who meet the inclusion criteria, will be primarily targeted and followed up until August 2024.

Study population

The study will begin recruitment using snowball convenience sampling in February 2024 and April 2024. Participants registered in cardiovascular medicine will be recruited and screened for eligibility at each site. After obtaining written consent following verbal and written explanations of the study, paper questionnaires will be distributed to the participants or QR codes will link them to the assessment form.

Inclusion criteria: age≥18 years; hospitalised due to a diagnosis of CHD; can be followed up by telephone; understand the purpose of the study; provide written informed consent.

Exclusion criteria: terminal or poor health status that prevents them from completing the questionnaire; inability to read, understand or complete questionnaires in Chinese language; a history of depression, anxiety or other mental disorders.

Follow-up and missing data

We will follow up with the patients via telephone or in-person interviews at 1, 3, and 6 months after collecting baseline measurements. Follow-up will include the measurement of anxiety and depression symptoms. The QOL will be measured only at 6-month follow-up (table 1). Because the ideal rate of missing data is usually <5%, we propose assessing a larger sample during the baseline period to ensure that ~500 individuals could participate in the 6-month follow-up. To further minimise attrition rates, participants will be reminded to provide not only their own but also the telephone numbers of family members during their first assessment. We will try to contact participants on different days or times if they cannot be reached by telephone. If necessary, patient updates will be determined from the electronic medical record system. Participants will be compensated for each assessment by random draws for ¥10. Multiple imputations will be applied to recalculate missing follow-up data if necessary.

Table 1.

Schedule of enrolment, assessments and outcome measures per follow-up

Study period
	Screening	Enrolment	Follow-up				Close-out
Timepoint	−t1	0	t0	t1	t2	t3	tx
Eligibility screen	X
Informed consent	X
Enrolment		X
Assessments
Demographic data
Gender			X
Age			X
Phone number			X
Smoking			X
Height and weight			X
PHQ-9
Anhedonia			X	X	X	X
Sad mood			X	X	X	X
Sleep			X	X	X	X
Energy			X	X	X	X
Appetite			X	X	X	X
Concentration			X	X	X	X
Motor			X	X	X	X
Guilt			X	X	X	X
Suicide thoughts.			X	X	X	X
GAD-7
Nervousness			X	X	X	X
Uncontrollable worry			X	X	X	X
Excessive worry			X	X	X	X
Trouble relaxing			X	X	X	X
Restlessness			X	X	X	X
Irritability			X	X	X	X
Feeling afraid			X	X	X	X
QOL
Self-evaluation of mental health						X
Satisfaction with social activities						X
Outcome measures
Network structure			X	X	X	X
Network centrality and connections			X	X	X	X
CIs			X	X	X	X
Correlation stability coefficient			X	X	X	X
Network comparison						X
Data analysis							X

GAD-7, Generalized Anxiety Disorder Scale; PHQ-9, Patient Health Questionnaire; QOL, WHO Quality of Life-brief version.

Data collection

All personnel will undergo research-related training before the study starts. Two personnel will conduct a pre-experiment using the questionnaire, and we will modify the questionnaire based on their feedback if necessary. For quality assurance, duplicates and questionnaires that are too short (<1 min) or too long (>1 hour) to complete will be excluded. Due to the nature of the study, participants and those directly involved cannot be blinded. However, different personnel will be assigned for data collection and analysis. The personnel responsible for data analysis will not directly participate in the survey investigation.

Data analysis

Descriptive analyses

All data will be statistically analysed using R V.4.2.1 (R Foundation for Statistical Computing, Vienna, Austria). Basic demographic characteristics and the severity of symptoms will be described using metrics such as frequencies, percentages, means and SD. We will process the data using packages in R V.4.2.1. Relevant packages will be used to help visualise the symptom network and understand its structures and relationships. The data analysis will comprise descriptive analyses of demographic data, network construction and estimation, accuracy and stability estimation, and comparisons of network characteristics at each timepoint.

Network estimation

We will collect various types of data as described.²⁷ We will estimate the network at each timepoint and construct a weight network using the Gaussian pairwise Markov random field graphical model.²⁴

Nodes represent symptoms, and edges or ties represent independent correlations between two nodes in symptom networks. The strength of the edges is operationalised as a partial correlation coefficient between nodes. We will use a weighted network to interpret the operationalisation of edge strength. To achieve a sparser visualisation and clearer interpretation, we will select the least absolute shrinkage and selection operator and extend the Bayesian information criteria to shrink edges in the network and select related tuning parameters. We will estimate and visually illustrate the network model using the qgraph and bootnet packages in R and generate an undirected association network using the Fruchterman-Reingold algorithm and spring layout. Covariates that affect the network will be considered for inclusion.

Network inference

To gain insight into the severity of items and discriminate between symptom clusters, the major indices of strength, betweenness and closeness will be used to quantify the most important items in the network. Bridge symptoms will be identified using the qgraph and networktools packages in R. The predictability for each node will be determined using the ‘flow’ function and the mgm package. This will identify items that are directly or indirectly connected with QOL.

Temporal network stability and bridge connection

We will compare more than two network structures that are still limited by statistical algorithms as described by Epskamp.³¹ We will combine structure diagrams from each timepoint to compare and describe structural differences and similarities using the NetworkComparisonTest (NCT) package. This package can assess differences between two or more networks over time. Demographic characteristics associated with depression and anxiety in patients with CHD will subgrouped for NCTs. The results will be expressed as p values. Furthermore, correlations between adjacency matrices and strength centrality will be estimated to determine relationships among networks.

Stability and accuracy analyses

We will apply the following measures to ensure the reliability of each timepoint symptom network and summarise the clinical application value. We will calculate CIs to estimate the stability of edge weights using non-parametric bootstrapping. In this computing environment, findings will be randomly resampled to generate new datasets within a 95% CI. The predicted influence and stability of the centrality indices will be assessed using correlation stability coefficients (CS-Cs). Generally, higher CS-Cs indicate a more stable network model, with CS-C>0.5 being considered applicable. The significance of differences between edge weights and centralities might be determined by bootstrapped difference tests using the bootnet package in R.

Study closure and participation suspended

Data collection will cease when the 6-month follow-up has been achieved and concluded for all participants who responded (excluding those lost to follow-up, or with missing data, or duplicates). Participants will have the right to withdraw from the study at any time without any responsibility or impact on their future medical care or eligibility for other studies.

Outcome measurements

We will collect baseline sociodemographic data and telephone number from the medical records of patients in the participating hospitals. The patients will self-report the sociodemographic variables of age, sex, height, weight and cigarette smoking habit. Height and weight will be measured in hospital. Clinical nurses will collect and verify all sociodemographic information on the first day of hospitalisation, then it will be entered into an electronic medical records system.

Measurement scales

Table 1 summarises outcome measures and scales that are all freely available in the public domain.

Patient Health Questionnaire (PHQ-9)

We will evaluate depression symptoms among patients with CHD using the Chinese version of PHQ-9.³⁴ Its items are based on the nine Diagnostic and Statistical Manual of Mental Disorders-IV criteria for major depressive disorder, namely anhedonia, sad mood, sleep, energy, appetite, guilt, concentration, motor symptoms and suicidal thoughts.³⁵ Each item is scored on a 4-point Likert scale, ranging from 0 (none) to 3 (almost every day); total scores range from 0 to 27, with higher total scores indicating more severe symptoms of depression. The Chinese version of the PHQ-9 has been validated in Chinese samples and PHQ-9 has been previously used in screening depression symptoms in cardiovascular patients efficiently.^{36 37}

Generalized Anxiety Disorder scale (GAD-7)

Given the popularity of the PHQ-9 for assessing and monitoring depression severity, a new seven-item anxiety scale similar to the PHQ-9 response set has been developed to facilitate screens for panic and anxiety symptoms.³⁸ The GAD-7 comprises seven items that measure anxiety symptoms, namely nervousness, uncontrollable worry, excessive worry, trouble relaxing, restlessness, irritability and fear using the following scale: 0 (none), 1 (several days), 2 (half the time) and 3 (almost all the time). The total score range is 0‒21 with higher scores indicating more severe symptoms of anxiety. Both English and Chinese version of GAD-7 have been validated.^{39 40} Furthermore, GAD-7 can stably measure anxiety associated with CHD (Cronbach’s alpha, 0.89; composite reliability index, 0.90).⁴¹

WHO Quality of Life-Brief version (WHOQOL-BREF)

We will assess the QOL using the global QOL scale, derived from the WHOQOL-BREF, which includes two items that measure overall QOL. The WHOQOL-BREF with the Chinese version is used to assess the subjective QOL of patients and the general public. It was developed under the leadership of WHO over the past four decades and is convenient to use and valid across cultures.^42–44 Higher scores indicate a better QOL. The Cronbach’s coefficient of WHOQOL-BREF has also presented well for each domain in cardiovascular patients.⁴⁵

Primary outcome

Network structure of anxiety and depression in patients with CHD, including:

1. The number of edges connected to each node.

2. Thickness of each edge, representing the strength of associations between nodes.

3. The direction of the correlations, where green edges represent positive correlations and red edges represent negative correlations.

4. Identification of particular symptoms that are directly or indirectly related to QOL.

Network centrality and connections

1. Node strength: an index used to quantify the number of direct connections between nodes.

2. Closeness: an index that calculates the extent to which a node is indirectly linked to other nodes.

3. Betweenness: an index used to determine the importance of a node in the average path between two other nodes.

4. Bridge expected influence: an index used to identify important nodes that bridge symptomatic connections.

Network comparison

1. Global structure: test for differences in the entire network structure.

2. Global strength: differences in edge weights of two networks.

3. Edge strength: differences in edge weights.

Secondary outcomes

1. CIs: these intervals provide a measure of the accuracy of edges weights.

2. CS-C: this coefficient assesses the expected influence and the stability of centrality indices.

Data management

All collected data will be managed using EpiData (V.3.1). Data will not be analysed until the research team confirms the accuracy and completeness of the data. Members of the research team who will be blinded to group allocation will handle data classification, analysis and visualisation of the network structure. All investigators will be responsible for ensuring confidentiality. The USC programme steering committee will supervise the study and organise meetings every 3 months to provide professional guidance. The core members of the research team will meet weekly to oversee daily management and track progress.

Confidentiality

The original data and relevant documents from the study will be securely stored on a password-protected computer within the data management software for at least 3 years. Supervisors are authorised to access the relevant data, and prior written approval from the sponsoring institution will be required before distribution to third parties.

Patient and public involvement

Patients were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Ethics and dissemination

The project conforms to the principles of the Declaration of Helsinki (2013 amendment) and all local ethical guidelines. The ethics committee of University of South China approved the study (Approval ID: 2023-USC-HL-414). The ethics boards at The First Affiliated Hospital of University of South China, The Second Affiliated Hospital of University of South China, Nanhua Affiliated Hospital of University of South China, Guangdong Provincial People’s Hospital and ShunDe Hospital Guangzhou University of Chinese Medicine also provided approval.

Discussion

Anxiety and depression in patients with CHD have long been the focus of psychosomatic medicine research. Previous evaluations of anxiety and depression in patients with CHD have primarily focused on symptom severity based on total scores in relevant scales.^{46 47} This approach might overlook interactions among mental symptoms represented by individual items, and potentially lead to bias in the results.²⁴ However, interactions between symptoms are common and fundamental phenomena in the occurrence and development of mental disorders, such as anxiety and depression.⁴⁸ This lack of precision could easily result in poor generalisation and targeting during intervention, and some reports have indicated that interventions cannot accurately target the exact symptoms, neglecting the role of each in the course of anxiety and depression.⁴⁹ The new concept of psychopathology research allows for exploration of the characteristics and mechanisms of anxiety and depression symptoms in the population with CHD from a comorbidity perspective.²⁴ Therefore, we conducted a cross-sectional and longitudinal study to compensate for the shortcoming of a single measure that cannot provide sufficient information about the disease course over time. This study protocol takes the lead in focusing on the value of the symptom network method for patients with CHD accompanied by anxiety and depression in the context of symptom interaction. Bridge symptoms might be intermediaries between symptoms and connections, and the centrality of symptoms could be targets for future intervention. Besides, dynamic network analysis at 1, 3 and 6 months will reflect differences between comparable network structures, determine the temporal stability of the network structure and identify the mechanisms of symptom changes over treatment cycles.

The potential limitations of our study are important to acknowledge. Since our sample will comprise hospitalised Chinese patients, the findings might not be generalisable to other regions or specific populations with anxiety and depression symptoms. Furthermore, our assessment of anxiety and depression symptoms will be based on self-report measures, which could result in recall bias and the Hawthorne Effect that might bias our findings to some extent. Besides, although this study will collect data at four timepoints, it might not fully explain the trajectory of the emergence, maintenance and progression of individual symptoms, such as a temporal network.

In summary, this multicentre study will aim to identify and measure the symptom network structure of depression and anxiety experienced by patients with CHD. The findings will contribute to a more integrated understanding of the comorbidity mechanism of care for patients with CHD accompanied by anxiety and depression. They will also provide meaningful information for clinical diagnosis and treatment.

Ethics statements

Patient consent for publication

Not applicable.