Network Analyses of the Symptoms in the Coronavirus Anxiety Scale (CAS) and its Associations with Depression, Anxiety, Stress and Alcohol Use


 Background

The Covid-19 Pandemic and subsequent actions taken by national/international organizations has generated a large amount of anxiety which may roam into the realm of pathology – COVID Anxiety. In order to measure this phenomenon, measures such as the CAS have been developed. The CAS being a self-report measure of anxiety-related physiologically symptoms that are aroused by information and thoughts related to COVID-19. However, as the CAS is fairly new tit requires validation and examination. This study fulfils this need through the use of Network Analysis.
Methods

The study used regularized partial correlation network analysis (EBICglasso) to examine the network structure of ratings of COVID anxiety symptoms as presented in the Coronavirus Anxiety Scale (CAS) and how these symptoms are related to distress (combination of depression, anxiety, stress) and alcohol use. A total of 968 adults from an Australian community sample completed the CAS, and measures of depression, anxiety, stress and alcohol use.
Results

The findings showed that the most central CAS symptom was abdominal distress, followed by tonic immobility. The symptom with the lowest strength centrality value was dizziness. Also, the network revealed at least moderate effect size connections for tonic immobility with dizziness, sleep disturbances abdominal distress, and for abdominal distress with appetite loss. Additionally, distress was associated positively with dizziness, tonic immobility and appetitive loss. Alcohol use was associated positively with dizziness and abdominal distress, and negatively with tonic immobility and appetitive loss.
Conclusions

Overall, the findings showed a novel understanding of the structure of the COVID anxiety symptoms in the CAS, and how these symptoms are associated with distress and alcohol use. The clinical implications of the findings for assessment and treatment of COVID anxiety and its comorbidity with distress and alcohol use are discussed.

appetite, and sleep ); cognitive (fear of virus casing preoccupation with threat provoking cognitions); and behavioral (preventive responses such as avoidance).
Evidence indicates that individuals prone to higher levels of COVID anxiety have a higher risk of developing severe mental health problems (Arora et al., 2020), including depression, stress, anxiety (Xiong et al., 2020; see also the review by Salari et al., 2020), and alcohol use (Gasteiger et al., 2021;Stanton et al., 2020), and that these may last well beyond the course of the pandemic (Nikčevića & Spada, 2020).
To date, several other measures have also been developed for measuring COVID-19 anxiety (and fear).  Lee, 2020a). A recent study showed that the FCV-19S, the FCQ and the FCQ scales/subscales measure different aspects of COVID anxiety and fear, thereby indicating that COVID anxiety is heterogeneous. Given these ndings, it can be argued that for a clear understanding of COVID anxiety, researchers need to identify what aspect of COVID anxiety their study is focusing on, and then consider this when interpreting their ndings (Mertens et al., 2020).

CAS
The CAS is a 5-item self-report measure of anxiety-related physiologically (somatic) symptoms that are aroused by information and thoughts related to COVID-19. The ve items of the CAS, all loading on a single factor, are (1) I felt dizzy, lightheaded, or faint, when I read or listened to news about the coronavirus (dizziness); (2) I had trouble falling or staying asleep because I was thinking about the coronavirus (sleep disturbances); (3) I felt paralyzed or frozen when I thought about or was exposed to information about the coronavirus (tonic immobility); (4) I lost interest in eating when I thought about or was exposed to information about the coronavirus (appetite loss); and (5) I felt nauseous or had stomach problems when I thought about or was exposed to information about the coronavirus (abdominal distress). According to Lee (2020b), the dizziness and tonic immobility symptoms in the CAS capture the physiological reactions of elevated fear to corona virus related stimuli. The sleep disturbances and appetite loss symptoms capture the physical effects of excessive worry about the coronavirus. The abdominal distress symptom captures fear and anxiety, result from a fearful reaction or the physical effect of excessive worry, or both.
Upon its construction, number of validation studies have examined the CAS, nding good psychometric properties. The initial CAS development and validation study supported a unidimensional structure, with high reliability (α = .93). Scores on the CAS correlated in theoretically meaningful ways with coronavirus diagnosis, functional impairment, coping through substance use and religion, hopelessness, and suicidal ideation. ROC analysis indicated 90% sensitivity and 85% speci city for detection and classi cation with a cutoff point ≥9 (Lee, 2020a). Based on a subsequent study (Lee, 2020b), the cutoff score was reduced to ≥5. To date numerous other studies have provided additional support for the psychometric properties of the CAS (e.g., Evren et al., 2020;Lee et al., 2020;Skalski et al., 2020). For example, Lee et al. (2020) found that dysfunctional scores on the CAS were associated with coronavirus infection, generalized anxiety, depression, functional impairment, perceived lack of social support, and suicidal ideation. For a full list of research studies that have used the CAS, the reader is referred to https://sites.google.com/cnu.edu/coronavirusanxietyproject/home.
Overall, therefore the CAS has sound psychometric properties. Although COVID anxiety is a heterogeneous construct, the CAS with its ve somatic items, is essentially a unidimensional, tapping physiologically and somatic anxiety and fear. As ready noted, despite it brevity and undimensionality the CAS is considered a valid measure for screening COVID 19 anxiety and has been used widely in COVID anxiety research. Therefore it can be argued that the CAS is ideally suited for examining the somaticrelated anxiety symptoms of COVID anxiety. Focusing on this group of symptoms is important as there is now some evidence that moderate to high levels of COVID-19 anxiety is associated with more somatic symptoms, even after controlling for generalized anxiety disorder (GAD), preexisting health problems, age, gender, and income.
A Novel Approach for Examining the Psychometric properties of the CAS To date, the psychometric properties of CAS have largely been examined from a latent variable perspective. In this perspective, it is assumed that there is a latent (unobservable) construct (which is the disorder/problem in question) that causes a range of observable responses (that are the symptoms of the latent disorder/problem). This is a re ective view of psychopathology. Seen in the context of COVID anxiety, the re ective view suggests that the COVID anxiety symptoms are responses arising from an assumed underlying latent COVID anxiety construct. This means that the COVID anxiety symptoms are interchangeable and equally re ective of latent COVID anxiety. Also, the COVID anxiety symptoms are considered to have nothing in common after controlling for the latent construct (an assumption referred to as local independence).
Although the latent variable approach (like that captured in a CFA) is currently the most dominant approach for understanding psychopathologies and related syndromes, a newly developed perspective, called the network approach, has a different view of psychopathologies. In the network framework, symptoms are understood as a causal system, interacting with each other in meaningful ways, resulting in the disorder or the syndrome (Borsboom & Cramer, 2013). A network model can be tested empirically using 'network analysis' (Borsboom & Cramer, 2013;Boschloo et al., 2015). Network analysis is an exploratory approach that provides visual and quantitative information about symptoms that are "core" or "central" (important) to the overall network of symptoms, and the strength of connections between symptoms (Borsboom & Cramer, 2013;Fried et al., 2015). As noted by Epskamp and others Epskamp et al., 2017), such a network can identify unique interactions between variables that cannot be identi ed using multiple regression analysis, and when the network analysis is exploratory it is advantageous over structural equation modelling (SEM), because there are no equivalent undirected models possible in SEM.

Clinical Importance of Network Analysis of the COVID Anxiety Symptoms
Results from network analysis of the symptoms of a disorder/syndrome can have important implications for theory, assessment and diagnosis, treatment and prevention. Traditionally, the theoretical importance of a symptom is viewed in terms of its severity which is ascertained in terms of its mean score. However, in network models, centrality, that is different from mean score, de nes the importance of a symptom. Indeed, the mean levels of symptoms can change without changes in their centrality in the network (Yang In relation to treatment, as symptoms for a disorder/syndrome identi ed as central in a network are considered most in uential in producing or maintaining the disorder/syndrome, intervening on these symptoms can be expected to maximize the impact of intervention. In this respect, and given its network characteristic, focusing on the central symptoms could potentially have a downstream effect in improving other network symptoms. Speci c to COVID anxiety, as note by Ramos-Vera (2021), network analysis allows clinicians to identify and understand more accurately the most important components in the dysfunctional dynamics of COVID anxiety, and consequently, ndings from network analysis, can contribute more effectively to detection and intervention of the negative effects of COVID-19.
Also, related to treatment, an expended network model that includes the COVID anxiety symptoms with other psychopathologies will increasing our understanding of the development and maintenance of other comorbid psychopathologies, which in turn would have major implications for preventing and treating these comorbidities.

Existing Network-based Psychometric Data for the CAS and for COVID Anxiety in General
To date, as far as we were able to establish, there have been only one study that has examined the network structure of COVID anxiety symptoms in the CAS (Ramos-Vera, 2021). The network was examined in terms of regularized partial correlation coe cients for ratings provided by a Peruvian community sample. The results indicate that appetite loss (symptom #4) was most central (indicating that it has greatest in uence in the network) and featured a strong connection (called edge weight) with dizziness (item # 1). Additionally, sleep disturbances (item #2) and tonic immobility (item # 3); and appetite loss (item # 4), and abdominal distress (item # 5) were also strongly connected with each other indicating a stronger in uence upon each other than they would have with less strongly connected Covoid-19, (4) traumatic stress symptoms associated with direct or vicarious traumatic exposure to COVID-19 and (5) COVID-19-related compulsive checking and reassurance-seeking. They reported three major clusters, with a cluster related toworries about the dangerousness of COVID-19 being most central, followed by the cluster related to the belief that the COVID-19 threat is exaggerated. The third cluster was related to compulsive checking, reassurance-seeking, and self-protective behaviors. Taken together, while these studies proved valuable information on the heterogeneous nature of anxiety and fear related to COVID-19, they offer no speci c network information on the CAS.

Limitations of Existing Network-Based Data on the CAS
Although there is some network analysis data on the COVID anxiety symptoms, as presented in the CAS, the ndings are limited. Firstly, there is only one study (Ramos-Vera, 2021). As network analysis is based on classical test theory, the ndings from such studies are largely sample dependent. Thus is need for replication studies. Secondly the only existing study that has used network analysis to examine the CAS (Ramos-Vera,2021), did not examine and report the accuracy and stability of the ndings for centrality and edge weights (connections between symptoms). This is a limitation as network analysis experts have recommended that a network must also be evaluated for its accuracy and stability . Thirdly, the Ramos-Vera (2017) study did not examine or illustrate how the network model would be used to examine how COVID anxiety contributes to the development and maintenance of other comorbid psychopathologies, which in turn would have major implications for understanding, preventing, and treating these comorbidities. For example, as already noted, pandemic-related psychological anxiety and distress are related to elevated levels of depression, stress, anxiety or distress (Xiong et al., 2020; see also the review by Salari et al., 2020), and alcohol use (Gasteiger et al., 2021;Stanton et al., 2020). From a network perspective, the inclusion of such distress and alcohol use together with the CAS symptoms in the same network model will reveal the speci c CAS symptom or symptoms that are central to the development and maintenance of these comorbidities, and therefore identify important targets of intervention. Fourthly, to date there is no network data for COVID anxiety symptoms in a Western community. Given these limitations, there is clearly need for more network analysis studies using Western samples and examining network ndings for accuracy and stability, and also relationships for CAS symptoms with potential comorbidities.

Aims of the Present Study
Given the limitations in existing network data on the CAS, and the positive clinical contributions that network analysis can offer, the major aim in the current study was to use network analysis, with regularized partial correlation, to examine the network structure of the ve COVID anxiety symptoms in the CAS (dizziness, sleep disturbances, tonic immobility, appetite loss, and abdominal distress) in a large Western (Australian) community sample. In the current study, we produced a network graph, displaying the topology of the symptom network, comprising the ve CAS somatic symptoms. We then evaluated statistically (using both edge width and centrality) the respective in uence of the symptoms in the network; and the robustness and stability of the network ndings. A secondary aim of the study was to compute an expended network model that included the COVID anxiety symptoms with distress and alcohol usage to ascertain the major associations of the COVID anxiety symptom with distress and alcohol usage.

Method Participants
Participants were 968 English speaking Australian adults from an online convenient sample from the general community. The age of participants ranged from 18 to 64 years (mean = 29.54 years; SD = 9.35 years), and included 622 men (64.3.7%; mean age = 29.46 years, SD = 8.93 years), and 315 women (32.5%; mean age = 30.02 years, SD = 10.39 years). Additionally, 26 individuals (2.7%) identi ed themselves as trans/non-binary gender, 1 individual identi ed as queer, and 4 individuals did not specify their gender. No signi cant age differences were found across men and women, t (935) = 0.846, p = .398. In terms of sociodemographic background, slightly more than half the number of participants reported being employed (55.0%) and most of them reported having completed at least secondary education (98.2%). Based on recommended scores of scores of ≥9 (Lee, 2020a) and ≥5 (Lee, 2020b), the number of individual screening positive for COVID anxiety were 35 (3.62%) and 123 (12.70%), respectively.

Measure
Coronavirus Anxiety Scale (CAS; Lee, 2020a) The CAS was used to measure COVID anxiety. It is a 5-item self-report measure of anxiety-related physiologically symptoms that are aroused by information and thoughts related to COVID-19. Participants rate each item in terms of how frequently they experience each anxiety symptom over the previous two weeks on a 5-point scale, ranging from 0 (Not at all) to 4 (Nearly every day over the last 2 weeks). Thus, higher scores indicate higher levels of a COVID-19 anxiety. The CAS has shown good reliability and validity (Lee et al., 2020). Although the Cronbach α for this scale in the current study was relatively low at .683, the McDonald omega coe cient was high at .87.

Depression Anxiety Stress Scales-21 (DASS-21; Lovibond & Lovibond, 1995)
The 21-item DASS-21 is a self-report measure with sub-scales for depression, anxiety, and stress. The anxiety items make no reference to COVID anxiety. All 21 items are rated on a 4-point scale (0 = did not apply to 3 = applied most of the time) in terms of how often the individual experienced the behavior during the past week. Although past evidence has shown acceptable convergent and discriminant validities, and high internal reliabilities for the DASS-21 sub-scales (Lovibond & Lovibond, 1995;Norton, 2007), more recent studies have supported a bifactor model, with a dominant general factor on which all the depression, anxiety and stress items load (e.g., Gomez et al., 2013). Thus, the DASS-21 items can be considered to measure general distress (Lee et al., 2019). The combined anxiety, depression and stress scale scores was used in the study to measure distress. The Cronbach's alpha for the full DASS-21 measure was 0.95 in the current study.
Alcohol Use Disorders Identi cation Test (AUDIT; Babor et al., 1992) Developed by the World Health Organization (WHO), the AUDIT is a 10-item self-report questionnaire with questions regarding amount and frequency of drinking, symptoms of alcohol dependence, and alcoholrelated problems. Each item is rated on a 4-point scale, ranging from 0 (never) to 4 (daily or almost daily). Thus, higher scores indicate more severity. The total score for the 10 items is generally used as an overall measure of the severity of hazardous or problematic drinking, and was used in the current study to measure alcohol addiction behavior. The AUDIT has good reliability and validity (Saunders et al., 1993). The internal reliability (Cronbach's alpha) for the total score for the current study was 0.89.

Procedure
This study was approved by the Human Ethics Research Committee, Victoria University (Australia). In order to gather participants, the study was advertised widely with Qualitrics links provided for participants to register their interest via social media (i.e. Facebook; Instagram; Twitter), the Victoria University websites and digital forums (i.e., reddit.com). The link took them to the Plain Language Information Statement (PLIS). Those wishing to participate were directed to click a button to agree to informed consent. This was followed by the questions seeking sociodemographic information, and the study questionnaires. Participants completed the online survey using a computer in a location of their choosing.

Statistical Network Analyses
In network analysis, variables are referred to as nodes, and the relationships between the nodes are referred to as edges. The strength of the relationship between nodes is indicated in terms of edge weights. Network nodes and edges can be estimated using zero-order correlations. In such instances, the edges between nodes will not control for the relations with other nodes, thus in ating correlations, and therefore resulting in di cult to interpret and misleading results. To overcome this, a regularized partial correlation approach, such as the graphical Least Absolute Shrinkage and Selection Operator (g-lasso; Tibshirani, 1996) is used to compute network analysis. Lasso shrinks small partial correlations to 0, resulting in a sparse network, and showing only the most important relationships in it. When a lassobased approach is applied, there is generally low likelihood of false positives, thereby providing con dence of edges reported in the network (Krämer et al., 2009). However, lasso can result in false negatives, and therefore the absence of an edge between two nodes cannot be automatically assumed to mean that there is no relation between them. Apart from visualization of the network graph, the network can be described statistically in terms of edge weights and centrality of the nodes (Borgatti, 2005). An edge weight indicates the strength of the relationship between nodes in terms of partial correlations coe cients. Centrality refers to the relative importance of the individual nodes in the network, i.e., a symptom with high centrality is one that is highly connected to other symptoms, and it may be the case that a central symptom is being activated by other symptoms. In contrast, a symptom with low centrality has fewer connections with other symptoms, and has less in uence on the network. Thus, while the network graph displays the topology of the symptom network, centrality indicates the relative importance of individual nodes within the network. Edge weights and the position of a node in the network determine its centrality.
Three commonly reported indices of centrality are betweenness, closeness, and degree (called strength in a weighted network, as is the case in the current study), (Opsahl et al. 2010). For reason that we will explain later, we used strength as our measure of centrality. In brief, strength is the sum of all direct associations a given symptom exhibits with all other nodes; and it re ects the direct in uence a given node has on the network. Nodes with high strength centrality values indicate that they are more central. Strength is known to re ect reasonably precise centrality estimates for psychology networks (Santos et al., 2018).
A network must also be evaluated for its accuracy and stability. A network's accuracy and stability refer to the likelihood that the network results will be replicated. For this, it has been recommended that the accuracy of the edge and the stability of the centrality estimates should be examined. One way to estimate the accuracy of edge weights is using bootstrap 95% non-parametric con dence intervals (CIs) . Narrower CIs suggest a more precise estimation of the edge . The stability of the centrality indices can be examined by using a different type of bootstrapping referred to as case-dropping (or alternatively node-dropping) bootstrapping ). This procedure examines if the order of centrality indices remains the same after re-estimating the network with less cases (or nodes). It quanti es the stability of centrality indices in terms of correlation stability coe cient. This coe cient re ects the correlation between the original centrality indices (based on the full data) and the correlation obtained from the subset of data representing different percentages of the overall sample. Although a correlation stability coe cient of 0.7 or higher has been suggested as being the threshold,  have suggested that the correlation stability coe cient should not be below 0.25, and preferably it should be above 0.5. For the current study, the stability of the centrality indices and edge accuracy of the network were examined using the procedures just described. Both were estimated with 1000 bootstrap samples.

Descriptive information of Data
There was no missing data. Initially we examined the mean and standard deviation (SD) scores. The ndings are presented in Table 1. As show, the mean score for the ve symptoms ranged from 0.23 to 0.48. The two most severely rated symptoms were sleep disturbances (2) and abdominal distress (5), and the symptom with the lowest severity was dizziness (1). Inspection of the distributions of the frequencies of the categories for each symptom indicated all ve response options for all ve symptoms were endorsed. Thus, it can be assumed that the ratings provided captured the full trait spectrum of the CAS symptoms.
Visualization of the COVID Anxiety Network Figure 1 shows a visualization of the network of the ve CAS symptoms. As shown, all symptoms were associated positively (blue edges) with one another, thereby indicating that all the symptoms were associated positively with each other.
Edge Weight of COVAD anxiety Symptoms in the COVAD Anxiety Network  (2) and abdominal distress (5).
The connection between abdominal distress (5) with appetite loss (4) of large effect size. The accuracy of the edge weights, estimated using bootstrap 95% non-parametric CIs is shown in Supplementary Figure S1. As shown, the CI ranges around all but four of the estimated edge-weights did not include zero thereby indicating fairly good precision for the edge weights. The CI of the four edges that included zero were dizziness-appetite loss, tonic immobility-appetite loss, dizziness-sleep disturbance, and dizziness-abdominal distress. Thus, caution is needed when interpreting these edges in the network.

Centrality of the Symptoms in the CAS Network
Prior to examining the centrality of the symptoms, we examined stability of the centrality indices for betweenness, closeness, and strength using case-dropping bootstrapping. The ndings are displayed in Supplementary Figure S2. The gure shows that for all centrality indices, the correlation stability (CS) coe cient from the subset of data representing different percentages of the overall sample. Supplementary Figure S2 shows that there was a slight drop in the correlations between the subsample estimate and the estimate from the original entire sample as the subset samples decreased from 95% of the original sample to 25% of the sample. However, for this, the correlations for the centrality indices for strength remained above .7 for decrease from 95% of to 25% of the sample, thereby indicating stability for the strength centrality indices . Given, this we examined centrality of the symptoms using only strength.
The standardized estimates of the centrality indices for strength are presented in Table 2. To ease interpretation, plots for the centrality measures in terms of z scores were created, and this is displayed in Fig. 2. For those interested, both Table 2 and Figure 2 also present the centrality indices for betweenness and closeness. As shown in Fig. 2 and Table 2, the two symptoms (in descending sequence) with the highest strength centrality values were abdominal distress (5) and tonic immobility (3). The symptom with the lowest strength centrality value was dizziness (1). These were also the case when the expected in uence centrality values were considered.

Network Analysis for the Associations of the CAS Symptoms with Distress and Alcohol Usage
In the network analysis model that examined the associations of the CAS Symptoms with distress and alcohol usage, the ve CAS symptoms, the DASS (Lovibond & Lovibond, 1995) distress scores, and the total sore for the AUDIT (Babor et al., 1992) were subjected simultaneously to network analysis. Table 3 shows the weights matrix from this analysis. As shown in this table, distress was associated positively with dizziness (1), tonic immobility (3) and appetitive loss (4). Alcohol use was associated positively with dizziness (1) and abdominal distress (5), and negatively with tonic immobility (3) and appetitive loss (4).

Discussion
The current study used network analysis to examine the structure of the ve COVID anxiety symptoms as measured by the CAS in a group of adults from the general Australian community. It examined the centrality of each COVID anxiety symptom in the network; the edge weights for the COVID anxiety symptom pairs, and the stability and accuracy of indices for centrality and edges. Additionally, the study examined how the COVID anxiety symptoms in the CAS were related to distress (combined depression, anxiety, and stress) and alcohol use.

COVID Anxiety Network Findings
Initially we examined the stability of the centrality indices , and in terms of strength discovered the following. The highest strength centrality values were abdominal distress, followed by tonic immobility. The symptom with the lowest strength centrality value was dizziness. It therefore follows that abdominal distress and tonic immobility may be the more important COVID anxiety symptoms, and dizziness may be the least important COVID anxiety symptoms. As there was support for the reliability for the strength centrality, these ndings can be considered as reliable.
The edge weight ndings showed that all connections were signi cant, thereby indicating associations with each other. Furthermore, the edge weights meeting at least the moderate effect size cut-off were for tonic immobility with dizziness, sleep disturbances and abdominal distress, and abdominal distress with appetite loss. As the test for the accuracy of the edge weights estimated using bootstrap 95% nonparametric CI did not include zero for these edges, there ndings can be considered as reliable.
Extended (CAS Symptoms with Anxiety, Depression, Stress, and Alcohol Usage) In the network analysis model that examined the associations of the CAS Symptoms with distress and alcohol usage, the ve CAS symptoms, the DASS (Lovibond & Lovibond, 1995) distress scores, and the total sore for the AUDIT (Babor et al., 1992) were subjected simultaneously to network analysis. Our ndings showed that distress was associated positively with dizziness, tonic immobility and appetitive loss. Alcohol use was associated positively with dizziness and abdominal distress, and negatively with tonic immobility and appetitive loss.

Comparison of Current and Past COVID Anxiety Network Findings
The ndings in the current study differ from the only previous study in this area by Ramos-Vera (2021).
Although that study also examined the network in terms of regularized partial correlation coe cients, it found, unlike the current study, that appetite loss was most central, and there were strong connection for appetite loss with dizziness, sleep disturbances and tonic immobility; and appetite loss and abdominal distress. Existing data show cultural and cross-national differences in the level of anxiety reported during . Given this, it is possible that such differences could explain (at least in part) the differences between our ndings and that reported by Ramos-Vera (2021). As mentioned previously, Ramos-Vera (2021) examined a Peruvian sample. In contrast, the current study examined an Australian sample. If so, it would imply that cultural factors may be important contributors to COVID anxiety network ndings and therefore COVID anxiety. At a more general level, this would mean that cultural factors need to be considered when treating COVID anxiety.
Another major difference is that the current study we examined how the COVID anxiety symptoms in the CAS were related to distress and alcohol use. Ramos-Vera (2021)

Novel Clinical Implications
Our ndings have novel implications for theory, classi cation, assessment and diagnosis, and treatment and prevention. We focus here on the major implications.
First, in a network, symptoms with high centrality values are considered most in uential in producing or maintaining the disorder/ syndrome. The highest strength centrality values were abdominal distress, followed by tonic immobility. Thus, it can be argued that the abdominal distress and tonic immobility symptoms are especially important for understanding and managing COVID anxiety, and therefore individuals with serious problems related to abdominal distress and tonic immobility are likely to demonstrate more serious COVID anxiety presentations. Therefore, clinicians may wish to pay special attention to the presence of these symptoms during assessment and diagnosis of COVID anxiety. As the symptom with the lowest strength centrality value was dizziness, it can be speculated that this symptom may not be critical for COVID anxiety.
Second, traditionally, the theoretical importance of a symptom is viewed in terms of its severity which is ascertained in terms of its mean score. The two most severely rated symptoms were sleep disturbances and abdominal distress. Given that in the network analysis, the highest two centrality symptoms were abdominal distress and tonic immobility, somewhat different conclusions about what are core symptoms in COVID anxiety were found when looking at symptom centrality and symptom severity (Mullarkey et al., 2019). Thus, it will be useful for clinicians to also consider symptom centrality when assessing and treating individuals with COVID anxiety.
Third, because the symptoms with high centrality values are considered most in uential, intervening on these symptoms can be expected to in uence other symptoms and in that way reduce the impact of the other symptoms also. This, therefore, could mean that focusing intervention efforts on abdominal distress and tonic immobility symptoms rather than the other symptoms could facilitate treatment effects. Where relevant, focusing on the symptoms with high centrality values (abdominal distress and tonic immobility) may also reduce the chances of on-set and development of COVAD anxiety in the context of primary prevention protocols implemented in the community.
Fourth, the edge weight ndings suggest that tonic immobility is more likely than the other symptoms to be associated with dizziness, sleep disturbances and abdominal distress, and also that abdominal distress is more likely than the other symptoms to be associated with appetite loss was of large. It would be useful for clinician to keep these associations in mind when assessing and treating individuals referred for COVID anxiety. Seen together with the ndings for centrality, it can be speculated that abdominal distress (feeling nauseous or having stomach problems when thinking about or being exposed to information about the coronavirus) and tonic immobility (feeling paralyzed or frozen when thinking about or being to information about the coronavirus) are probably the symptoms most in uential in COVID anxiety. According to Lee (2020b), the tonic immobility symptoms in the CAS capture the physiological reactions of elevated fear to corona virus related stimuli; and the abdominal distress symptom captures fear and anxiety, result from a fearful reaction or the physical effect of excessive worry, or both.
Fifth, the extended network model showed that the dizziness symptom was associated positively with distress and alcohol use; the abdominal distress symptom was associated positively with alcohol use; and the tonic immobility and appetite loss symptoms were associated positively with distress. This indicates that higher levels of the dizziness symptom will be associated with distress and alcohol use, higher levels the abdominal distress symptom will be associated with alcohol use; and higher levels of tonic immobility and appetite loss symptoms will be associated more distress. It would be useful for clinician to keep these associations in mind when assessing and treating individuals referred for COVID anxiety.

Limitations and Directions for Further Studies
Despite the positive value of the ndings in the current study, the results in the study have to be interpreted in the light of a number of limitations. Firstly, network analysis assumes that mental disorders (and therefore COVID anxiety) are causal systems. However, as we used cross-sectional data in the current study, causality cannot be securely assumed. At best, we were able to eliminate spurious candidates for causal relations. Causality assessment would require longitudinal data, collected repeatedly. Further studies may wish to examine such concerns, using longitudinal network analysis.
Secondly, as we conducted the network analysis using a normative-community sample, the ndings cannot be directly generalized to other samples, like speci c racial and clinical groups. Thirdly, as we used a self-rating measure of COVID anxiety, the ndings may not be applicable to data collected via clinical interviews, or from other sources. Fourthly, as our ndings are based on group-level analyses, it may not be directly applicable at the individual level. It is possible that some of the associations found in the current study may not be applicable to some individuals. Clearly, we need more network studies of the COVID anxiety symptoms, using longitudinal data, collected using multiple sources and methods and different racial and clinical groups. Individualized networks would also be bene cial for a comprehensive understanding of the COVID anxiety network. Despite these limitations, our ndings do offer novel insights on the structure of COVID anxiety symptoms, and their relative importance that can be used effectively for theorizing, assessing and treating COVID anxiety.     Figure 1 Network of the CAS COVID Anxiety Symptoms