Role of interoceptive fear and maladaptive attention and behaviors in the escalation of psychopathology—a network analysis

ABSTRACT

The complex interplay of fear, attention, and behavior toward bodily sensations with psychopathological symptoms and how they mutually influence and potentially reinforce one another remains to be fully elucidated. In this study, we used a network analytical approach to unravel these complex interactions. Specifically, we aimed to identify central symptoms and etiologically relevant factors that might be associated with anxiety and depressive core symptoms. To this end, the following clusters were assessed in 791 adults: interoceptive fear, interoceptive attention, maladaptive behaviors related to bodily sensations, and core symptoms of anxiety and depression. This network was modeled using a Gaussian Graphical Model. Central variables (nodes) were identified using centrality indices and bridge analysis. Self-examination and attention to bodily sensations emerged as central nodes. Moreover, time spent paying attention to bodily sensations, fear of anxiety-related sensations, and self-examination were identified as central bridge nodes, that is, central nodes connecting psychopathologically relevant symptom clusters. The present study indicates that fear of bodily sensations, the amount of attention and time spent focusing on somatic sensations, and self-examination are central factors. The findings suggest potential targets for future longitudinal studies on the impact of these factors for the escalation of anxiety and depressive symptoms.

Highlights

• Central variables were identified through centrality indices and bridge analysis

• Attention to bodily sensations and self-examination were identified as central nodes

• Fear of bodily sensations and self-examination emerged as central bridge nodes

• Time spent paying attention to body sensations also emerged as central bridge node

• Results suggest possible targets for future experimental and longitudinal research

KEYWORDS:

• Interoception
• psychopathological network
• anxiety
• depression
• COVID-19

Introduction

Perception, processing, and responding to bodily sensations are interconnected mechanisms that exhibit mutual influence and reinforcement. Alterations of these mechanisms can impact psychopathology, in particular anxiety and mood disorders (Khalsa et al., 2018; Paulus et al., 2019). These altered mechanisms related to internal body states may manifest in emotional (i.e. increased fear of bodily sensations), attentional (e.g. increased sensibility or attention toward bodily sensations), and behavioral alterations (e.g. increased avoidance and safety behavior) which may result in functional and social impairments in everyday life, as well as the development and maintenance of psychopathology (Hartmann et al., 2019; van Diest, 2019). While existing research has examined isolated interactions within each of these factors as outlined below, the complex mutual influence and potential reinforcement among them remains elusive. Therefore, it is necessary to integrate these key factors within a comprehensive framework to extend our knowledge of their relation to psychopathological symptoms. Network analysis may help to provide new insights into unravelling these complex interactions (Borsboom, 2017). Understanding the interdependence of these individual mechanisms and symptoms will provide first indications to inform future experimental and longitudinal studies, thus helping to refine etiological models and develop more efficient and effective interventions.

Previous studies have emphasized anxiety sensitivity, i.e. fear of bodily sensations due to concerns about harmful consequences (e.g. a potential heart attack) of such sensations, as a transdiagnostic risk factor associated with several psychopathological symptoms, such as anxiety and depression (Naragon-Gainey, 2010; Schmidt et al., 2006). In particular, when individuals are confronted with body symptoms during provocation challenges (e.g. guided hyperventilation or loaded breathing), fear of bodily sensations has been demonstrated to be associated with increased symptom reports and exaggerated defensive mobilization (Benke et al., 2019; Brown et al., 2003; Melzig et al., 2011). Moreover, there is convergent evidence that increased fear of bodily sensations is also linked to maladaptive attentional and behavioral changes towards the feared body sensations (Fergus & Bardeen, 2013; Zvolensky & Forsyth, 2002). Avoidance and safety behaviors that target at preventing feared consequences related to the occurrence of bodily sensations (e.g. a panic attack, or a potential heart attack, Bouton et al., 2001; Dixon et al., 2013) may include avoidance of specific places or activities (e.g. climbing stairs, drinking caffeinated beverages) as well as body-checking, reassurance or information-seeking behavior (Hartmann et al., 2019; Helbig-Lang & Petermann, 2010). These behaviors are well known to play a central role in maintaining threat beliefs, anxiety symptoms, and fear of bodily sensations by preventing the disconfirmation of concerns about the consequences of bodily sensations as has been evidenced by a number of longitudinal and experimental studies (Benke et al., 2019; Wilson & Hayward, 2006).

Fear of bodily sensations, however, is thought to increase the likelihood of engaging in avoidance and safety behaviors by amplifying the expected or experienced aversiveness of somatic sensations (Horenstein et al., 2018). Likewise, increased attention allocation or hypervigilance to bodily sensations has been shown to be associated with a higher risk of engaging in safety and avoidance behaviors to prevent, alleviate or stop the experience of bodily sensations (Abramowitz & Moore, 2007; Olatunji et al., 2007). It is suggested that increased interoceptive attention increases the likelihood of detecting bodily sensations and increasing the number of such experienced sensations, with catastrophizing interpretations of these sensations (i.e. interpretation of bodily sensations as potentially dangerous) and amplifying fear (Asmundson et al., 2010; Leonidou & Panayiotou, 2018). In this regard, it is not surprising that individuals with heightened fear of somatic sensations are characterized by an increased level of interoceptive attention, as indicated by an increased attentional focus to bodily sensations, body sensitivity, amount of attention, and time spent attending to bodily sensations (Barsky & Wyshak, 1990; Domschke et al., 2010).

Experiencing anxiety symptoms and heightened arousal as uncontrollable may further lead to perceived helplessness. Individuals who tend to attribute such events as uncontrollable and suffer from anxiety disorders may have a heightened risk to develop depressive symptoms (Heimberg et al., 1989; Mineka et al., 1998). Additionally, previous research has documented high comorbidity between anxiety and depressive disorders (Lamers et al., 2011; McGrath et al., 2020). Furthermore, attentional focus to bodily sensations proved to be a significant predictor of severe symptoms of general depressive symptomatology, after controlling for relevant influencing factors (trait anxiety, fear of bodily sensations and presence of health issues, McLaughlin et al., 2016).

In light of these previous studies, it becomes clear that fear, attention and behavior toward bodily sensations are centrally linked not only to each other but also to anxiety and depressive symptoms, which may entail an escalating process, thus increasing the risk of mental health sequelae. While numerous studies have examined these constructs individually, a significant gap remains in our understanding of their interplay and how they potentially potentiate each other. Yet, the importance of separate symptoms and transdiagnostic factors such as fear of bodily sensations, attentional sensitivity to bodily sensations, information-seeking and reassurance behavior, or feeling nervous and anxious as well as their interactions, in escalating psychopathology remains uncovered. To elucidate these complex interactions, network analytical approaches have been proposed. In network analysis, the associations between co-occurring symptoms or variables (referred to as nodes) and their relationships (edges as the links between the nodes) to each other are examined (Borsboom, 2017; McNally, 2016). The network approach not only allows to investigate and identify associations between symptoms and signs of mental diseases but also provides the opportunity to identify initial indications of potential central symptoms, which must then be examined in longitudinal studies to understand their role in the development of psychopathology, thereby potentially informing the development of more effective therapeutic interventions (Borsboom & Cramer, 2013; Borsboom et al., 2021).

In the present study, we used a network analytical approach to attain a comprehensive understanding of the complex interactions between fear of bodily sensations, body vigilance and maladaptive behavior, as well as core symptoms of anxiety and depression. The present study was conducted during the COVID-19 pandemic when individuals were faced with a potential threat to their health. Based on evidence from previous epidemics (e.g. H1N1 influenza, see Wheaton et al., 2012; Ebola virus, see; Blakey et al., 2015; Zika virus, see Blakey & Abramowitz, 2017), we included fear and attentional processes related to COVID-19 to test whether they might influence this network consisting of transdiagnostic signs and core symptoms of anxiety and depression. Specifically, we aimed to identify the central components that might be associated with other symptoms (i.e. central symptoms and bridge symptoms), which could then potentially be targeted in future longitudinal and experimental studies.

Methods

Participants

A cross-sectional study was conducted between 29 April 2020, and 4 June 2020. Convenience sampling methods (e.g. mail lists and social media) were used to recruit participants. The survey was completed online (at socisurvey.de). In total, 836 participants participated in the online survey. The following exclusion criteria were applied: less than 500 seconds spent on the questionnaire (n = 23) to ensure completion, participants who identified as diverse gender due to their limited representation in the dataset (n = 5) to avoid potentially unreliable or statistically unstable conclusion, and the presence of previous or current COVID-19 infection (n = 18) due to its possible influence on neurological and psychological processes (Koralnik & Tyler, 2020; Serafini et al., 2020). Overall, data from 791 persons entered the analysis. The characteristics of the present study sample are summarized in Supplemental Table S1. As shown in Supplemental Table S1, the mean age of the sample was 46.55 years (SD = 14.35, range: 18 to 91 years), 80.3% (n = 635) were female, 46.6% (n = 369) had a college/university degree. More than half of the sample (n = 452, 57.1%) considered themselves to belong to an official risk group for a severe course of disease following a COVID-19 infection (Rommel et al., 2021). Each participant provided informed consent. Approval for the study was provided by the local Ethics Committee of the University of Marburg.

Measures

Interoceptive fear

Fear of contracting COVID-19

Participants were asked to rate their level of fear of contracting COVID-19 on one item ranging from 0 to 100 (no anxiety to very high anxiety).

Anxiety sensitivity scale

The physical concerns subscale (PC) of the German version of the Anxiety Sensitivity Index-3 (ASI-3; Kemper et al., 2011) was used to measure concerns regarding harmful physical consequences of anxiety-related sensations (Taylor et al., 2007). The physical concerns subscale comprises six items (e.g. “It scares me when my heart beats rapidly”) rated on a five-point Likert scale ranging from 0 (“does not apply at all”) to 4 (“fully applies”). The ASI-3 has demonstrated good validity (Taylor et al., 2007) and excellent internal consistency (Cronbach’s α) (α = .90).

Interoceptive attention

The first three items of the Body Vigilance Scale (BVS, Schmidt et al., 1997) were utilized to assess the general level of attentional focus to bodily sensations, sensitivity to bodily sensations, and average time spent attending to bodily sensations during the last two weeks. Moreover, participants were asked to rate the amount of attention they pay to sensations related to anxiety like sensations of exertion, to the abdomen and chest, and other sensations such as numbness and tingling in the last two weeks. An additional item measured the amount of attention paid to body symptoms related to COVID-19 in the last two weeks. All items were rated on an 11-point Likert scale. Previous studies have demonstrated good internal consistency and test-retest reliability (Olatunji et al., 2007; Schmidt et al., 1997). Internal consistency in the current sample was excellent (α = .91).

Safety/Avoidance behavior

Safety-seeking behaviors

The reassurance behavior subscale of the Hypochondriac Safety Behavior Questionnaire (Weck et al., 2012) was utilized to assess the level of safety behaviors used during the last seven days. The reassurance behavior subscale (Weck et al., 2012) assesses self-examination (e.g. “Do you check your body for abnormalities?”), and information-seeking behavior (e.g. “Do you ask your physician about serious illnesses?”) with four items each. All items were rated on a five-point Likert scale (0 = “never” to 4 = “very often”). The FSVH demonstrated good convergent and discriminant validity. Similar to previous research (Weck et al., 2012), our study demonstrated good internal consistency (α = .85).

Avoidance behaviors

Behavioral avoidance of somatic symptoms (i.e. stress and somatic symptom avoidance behavior) was assessed using items of the Texas Safety Maneuver Scale (TSMS, Helbig-Lang et al., 2014; Kamphuis & Telch, 1998) questionnaire rated on a five-point scale (0 = “never” to 4 = “always”). In the present study, we only considered items of the stress and somatic symptom avoidance behavior subscales including situations that were unaffected by pandemic-related restrictions (i.e, eight items; see Supplemental Table S2). An additional item (“climbing stairs”; generated by the authors) was added to the scale. For the present sample, there was excellent internal consistency (α = .91).

Core symptoms of anxiety and depression

The generalized anxiety disorder scale-2

Anxiety symptoms (i.e, feeling nervous, anxious, or on edge, and not being able to stop or control worrying) were assessed by the Generalized Anxiety Disorder Scale-2 (GAD-2, Plummer et al., 2016; Spitzer et al., 2006) with two items rated on a four-point Likert scale (0 = “not at all” to 3= “nearly every day”). Cronbach´s alpha in the present sample was good (α = .84).

The patient health questionnaire-2

Depressive symptoms (depressed mood, anhedonia) were assessed via the Patient Health Questionnaire-2 (PHQ-2, Löwe et al., 2005) with two items rated on a four-point scale (0 = “not at all” to 3 = “nearly every day”). Internal consistency in the current study was good (α = .84).

Data analysis

Preprocessing and descriptive analyses were performed using the statistical program IBM SPSS Statistics (Version 27, for Windows). Network analyses were calculated using R (Version 4.2.2). For network analysis, the packages bootnet (Version 1.5, Epskamp et al., 2018), networktools (Version 1.1.0, Jones, 2017), and qgraph (Version 1.9.2 Epskamp et al., 2012) were employed.

Network analysis

In network analyses, nodes represent psychological variables and edges represent the partial correlations between these variables (Belvederi Murri et al., 2020; van Borkulo et al., 2015). Based on theoretical assumptions, we pre-defined four communities: interoceptive fear, interoceptive attention, safety/avoidance behavior, and anxiety/depression. An overview of the communities and single nodes is shown in Figure 1 (see also Supplemental Table S2 for a description of the single nodes of the network). In the present study, we estimated a partial correlation network in which edges between two nodes were calculated while controlling for the impact of all other network nodes (Epskamp & Fried, 2018). A detailed description can be found in the Supplemental Material (see S 2.3.1.).

Figure 1. Regularized partial correlation network of the single nodes and bridge symptoms in healthy individuals.

Correlations smaller than .01 are not shown to enhance interpretability. The nodes are colored according to the pre-defined communities and bridge symptoms. The green edges indicate positive regularized partial correlations.

Node centrality

Centrality indices (strength, closeness, and betweenness) were analyzed (Borsboom, 2017; Opsahl et al., 2010). A detailed description can be found in the Supplemental Material (see S 2.3.1.1). Strength centrality showed the highest stability (see analysis section and Supplemental Fig. S1) and was therefore utilized in the present study.

Bridge analysis

To analyze the most important nodes connecting individual communities (i.e. 1. Interoceptive fear, 2. Interoceptive attention, 3. Safety/avoidance, 4. Anxiety/depression), bridge analysis was performed (Jones et al., 2021). Bridge symptoms that most strongly connected different communities were identified via bridge centrality and analyzed using the package networktools in R (Jones, 2017). A detailed description can be found in the Supplemental Material (see S 2.3.1.2). In the present study, bridge strength centrality showed the highest stability (see analysis section and Supplemental Fig. S1) and was therefore utilized in the present study. To prevent confirmation biases in our interpretation of the bridge centrality data, we selected the bridge nodes based on a blind 80^th percentile cutoff on the bridge strength scores (Jones et al., 2021).

Accuracy and stability

After performing network analysis, bootstrapping methods (2500 iterations) with the package bootnet (Epskamp et al., 2018) were utilized to estimate the stability and accuracy of the network (Chernick, 2008; Efron, 1979; Epskamp & Fried, 2018). A detailed description can be found in the Supplemental Material (see S 2.3.1.3). The replicability of the network was analyzed through simulation studies using the bootnet package in R (Epskamp et al., 2018).

Results

Descriptives and zero-order correlations

The descriptive statistics of all variables are presented in Supplemental Table S3. All variables showed weak-to-moderate positive associations with each other (all p-values < .01, see Supplemental Table S4).

Network estimation

The regularized partial correlation network is shown in Figure 1. In total, 59 of the 91 possible edges were non-zero edges in the network. The edge weights in this network ranged from −0.06 to 0.68 (see Supplemental Table S5 for all individual edge weights). The strongest associations were found between the variables within the same community. In this network, attention to bodily sensations (BVS, node 3)—interoceptive sensitivity (BVS, node 4, edge weight = 0.68), anhedonia (PHQ-2, node 11)—depressed mood (PHQ-2, node 12, edge weight = 0.49), information seeking (FSVH, node 8)—self-examination (FSVH, node 9, edge weight = 0.47), and depressed mood (PHQ-2, node 12)—worrying (GAD-2, node 14, edge weight = 0.41) represented the strongest edges (see Supplemental Fig. S2, bootstrapped CIs of these edges did not overlap with any other bootstrapped CIs of other edges), which was substantiated by significant difference tests between edge weights (see Supplemental Fig. S3).

However, as shown in Figure 1, variables were also associated with variables of other communities, but less strongly than within the same community (see also Supplemental Table S5 and Supplemental Fig. S3). Depressed mood and anhedonia showed only a few negative associations with the variables of other communities (i.e. interoceptive fear, interoceptive attention, and safety/avoidance). In contrast, anxiety symptoms displayed numerous positive associations with variables of other communities (e.g. feeling nervous or anxious [node 13] was positively associated with fear of COVID-19 [node 1, edge weight = 0.10], fear of bodily sensations [node 2, edge weight = 0.08], attention to anxiety symptoms [node 7, edge weight = 0.09], and avoidance behavior [node 10, edge weight = 0.02]). Fear of bodily sensations (node 2) showed positive edges to all other variables (except for core depressive symptoms). Fear of COVID-19 (node 1) showed positive edges to fear of bodily sensations (node 2, edge weight = 0.17), time spent attending to body symptoms (node 5, edge weight = 0.08), attention to COVID-19-related body symptoms (node 6, edge weight = 0.14), information seeking (node 8, edge weight = 0.12), and avoidance behavior (node 10, edge weight = 0.06). Self-examination (node 9) exhibited stronger associations with time spent attending to bodily sensations (node 5, edge weight = 0.13) and fear of bodily sensations (node 2, edge weight = 0.12) but no associations with depressive (nodes 11 and 12) and anxiety core symptoms (nodes 13 and 14). Moreover, attention to anxiety-related sensations (node 7) was associated with fear of bodily sensations (node 2, edge weight = 0.10), self-examination (node 9, edge weight = 0.13), avoidance (node 10, edge weight = 0.10), and feeling nervous or anxious (node 13, node = 0.09).

Centrality

The z-standardized indices of strength, betweenness, and closeness centrality are shown in Supplemental Fig. S4 (Supplemental Table S6 for centrality indices, for strength centrality see Figure 2, Supplemental Fig. S5 for raw strength centrality). The nodes showed considerable variation in their centrality. The strongest strength centrality was observed for feeling down, depressed, or hopeless (node 12, 12.52), followed by attention to anxiety-related sensations (node 7, 10.97) and self-examination (node 9, 10.88). These nodes also proved to be significantly stronger in terms of strength centrality than half of the other nodes (see Supplemental Fig. S3 for the results of the nonparametric bootstrapped difference tests). Avoidance of somatic sensations (Node 10, 0.61) showed the lowest strength centrality (see Supplemental Table S6 for the raw values of centrality indices).

Figure 2. Z-Standardized centrality and bridge centrality indices of the single nodes from the regularized partial correlation network in healthy individuals.

Z-Standardized strength centrality (left panel) and bridge centrality indices (right panel) corresponding to the regularized partial correlation network in healthy individuals; Each point on the y-axis represents a node in the network.

Bridge analysis

To identify the variables most connected to the variables of the different communities (see section Network Estimation), a bridge analysis was conducted. As shown in Figure 2, time spent attending to bodily sensations (node 5, bridge strength centrality index = 0.061), fear of bodily sensations (node 2, bridge strength centrality index = 0.060) and self-examination (node 9, bridge strength centrality index = 0.052) emerged as the three variables with the highest bridge strength centrality (see Supplemental Table S6 for all individual bridge strength centrality indices). The values of the other bridge centrality indices (bridge betweenness, bridge closeness) are presented in the Supplemental Material (see Supplemental Table S6 for raw bridge centrality values, Supplemental Fig. S6 for z-standardized bridge centrality indices, and Supplemental Fig. S7 for raw bridge strength centrality).

Network stability and accuracy

The stability of the edge weights is shown in Supplemental Fig. S2. As can be seen, the confidence intervals were relatively narrow, implying stable outcomes. According to the case-dropping bootstrapping, strength centrality, and bridge strength centrality showed the highest stability (CS(cor = 0.7) = 0.75), indicating that with decreasing sample size (75%), the results are comparable to the primary findings (see Figure 3). The bootstrapping results for the other centrality and bridge centrality indices are shown in the Supplemental Material (Supplemental Fig. S3, and see Supplemental Table S6 for the centrality indices of each node). According to bootstrap difference tests (see Supplemental Fig. S3), most of the edge weights and strength centrality indices showed significant differences. The results of the simulation studies are presented in the Supplemental Material (see S 2.1 Simulation Studies).

Figure 3. Stability of strength centrality and bridge strength centrality indices using case-dropping bootstrap.

Correlation between the strength centrality and bridge strength centrality indices of networks sampled with persons dropped and the original sample. Lines display the means and areas show the range from the 2.5^th to the 97.5^th quantile.

Discussion

In the present study, we examined the complex interplay and importance of symptoms and transdiagnostic factors regarding interoceptive attention, fear of bodily sensations, and maladaptive behaviors to investigate their potential relevance in the escalation of core symptoms of anxiety and depression. Therefore, we utilized network analysis to characterize the associations between these factors and psychopathological symptoms of anxiety and depression. In the underlying network, the strongest associations were identified within the respective communities (i.e. interoceptive attention, interoceptive fear, maladaptive behavior, and core symptoms of anxiety/depression). Time spent scanning the body, fear of bodily sensations, and self-examination emerged as central bridge nodes, i.e. central factors that are associated with two or more symptom dimensions or clusters (e.g. maladaptive attention or behavior toward bodily sensations). Depressed mood, attention to anxiety-related sensations, and self-examination emerged as the nodes with the highest centrality, suggesting that these factors have the highest interrelation with other factors and therefore might contribute to symptom escalation, warranting further investigation in longitudinal studies to understand their potential role in symptom dynamics.

Previous studies demonstrated that the general tendency to attend to bodily sensations is associated with fear of bodily sensations, health anxiety, as well as safety and avoidance behavior, especially in patients with panic or illness anxiety disorder (Fergus & Bardeen, 2013; Hamm et al., 2014; Olatunji et al., 2007). The current study extends this evidence in demonstrating that specific attentional alterations (i.e. longer duration and greater severity of attention allocation to bodily sensations) are associated with increased anxiety core symptoms, and fear of bodily sensations as well as maladaptive behavior. Specifically, attention to anxiety-related sensations, such as sensations of exertion or numbness, proved to be a central factor in the present network and was mainly associated with other indicators of increased interoceptive attention, but also with self-examination, fear of bodily sensations, and the feeling of being nervous, anxious, or on the edge. Moreover, the time spent scanning the body for anxiety-related sensations (e.g. sweating, heart palpitations, and dizziness) emerged as the factor with the strongest bridge strength, with associations to fear of bodily sensations, avoidance, safety behaviors, and worrying. These findings could suggest that these specific alterations in attention toward anxiety-related sensations may be linked to the proliferation of anxiety and depressive symptoms by influencing fear of bodily sensations, maladaptive behavior, and anxiety symptoms. However, the direction of these associations ought to be probed in future longitudinal and experimental studies. Also, future research should aim to explore whether this alteration in attentional processes could potentially serve as a transdiagnostic causal factor or if it is primarily associated with specific disorders such as panic disorder or illness anxiety disorder (Asmundson et al., 2010; Ehlers & Margraf, 1989; Warwick & Salkovskis, 1990). Although the results outline possible targets for future research caution is advised for causal conclusions from our present cross-sectional network models.

Self-examination emerged as a factor with high centrality and bridge strength in the present network, indicating that, among other maladaptive behaviors, particularly self-examination is associated with fear and altered attention processes to bodily sensations. While self-examination has no substantial associations with depressive and anxiety core symptoms, it might exert its influence indirectly by increasing fear and attention toward bodily sensations. This highlights its potential significance within the network and underscores the need for further research to elucidate these complex associations. The present finding corresponds with previous studies in anxious populations, as well as current etiological models of anxiety pathologies, emphasizing safety and avoidance behavior as crucial factors for the maintenance of threat beliefs and fear (Abramowitz & Moore, 2007; Olatunji et al., 2011). In the context of fear of bodily sensations, reducing self-examination might be essential to allow the disconfirmation of interoceptive threat beliefs to boost inhibitory learning—a hypothesis that has to be tested in future interventional studies.

Furthermore, fear of bodily sensations emerged as a bridge node that exhibited connections to time spent scanning the body, attention to anxiety-related sensations, avoidance behaviors, and anxiety core symptoms. This is consistent with previous studies, where fear of bodily sensations was associated with avoidance and fear responses upon confrontation with symptom provocation challenges (e.g. hyperventilation), increased vigilance toward bodily sensations as well as a higher risk to develop anxiety symptoms (Benke et al., 2019; Melzig et al., 2008; Zvolensky & Forsyth, 2002). Thus, our data suggest that increased fear of bodily sensations potentially is tied to alterations in a broad set of symptom dimensions. Interacting with increased interoceptive attention (particularly attention to anxiety-related sensations and longer time spent monitoring bodily sensations), avoidance and safety behavior (particularly self-examination), fear of bodily sensations may set the stage for a vicious cycle responsible for the pathogenesis and perpetuation of psychopathology by increasing the salience of bodily sensations (Allan et al., 2014; Dixon et al., 2013). Even though the results indicate relevant associations between these factors, further experimental and longitudinal network studies are warranted to draw causal conclusions (McNally, 2021).

Among the psychopathological symptoms, the strongest associations with fear, attention, and behavior toward bodily sensations were observed for anxiety core symptoms, while there were only a few marginal associations with depressive core symptoms, supporting the role of fear, attention, and behavior toward bodily sensations in anxiety symptomatology, as highlighted in etiological models of anxiety pathologies (Hamm et al., 2014; Salkovskis et al., 1999). Consistent with previous network analyses of anxiety and depressive symptoms, depressive mood emerged as a central variable in the present network, which primarily resulted from the strong associations between depressive mood and anhedonia, as well as anxiety core symptoms, especially with uncontrolled worrying (Jones et al., 2021; Ren et al., 2021).

Fear of COVID-19 was primarily associated with fear of bodily sensations and attention to COVID-19 symptoms. This finding replicates studies that demonstrated that fear of COVID-19 is associated with higher fear of bodily sensations (Ojalehto et al., 2021; Rogers et al., 2021; Warren et al., 2021). While previous studies have revealed an association between fear of COVID-19 and interoceptive sensitivity (Elliott & Pfeifer, 2022; Suzuki et al., 2021; Zvolensky et al., 2022), the present data suggest that fear of COVID-19 is related to a specific attention allocation toward COVID-19 symptoms rather than a tendency to detect possible harmful bodily sensations in general. Moreover, in the present network, fear of COVID-19 displayed a rather low centrality, suggesting that more general or context-independent symptoms have a stronger impact on the overall symptom escalation than specific fear related to the COVID-19 pandemic. Furthermore, the timing of the survey may have contributed to the limited impact of fear of COVID-19 in this study (Bendau et al., 2022; Mauz et al., 2023; Robinson et al., 2022). Our data collection coincided with a summer plateau in COVID-19 cases, marked by a significant reduction in new infections (World Health Organization [WHO], 2023). This likely led to lower fear and perceived threat of COVID-19 among the population, influencing our findings related to fear of COVID-19 and thus should be considered when interpreting the present findings.

The present study indicates that fear of somatic sensations, time spent on body scanning, and self-examination might play critical roles in the illustrated symptom network. Thus, these factors might promote the escalation of symptoms by mutually influencing and augmenting other symptoms within the network, which could increase the risk of the manifestation and persistence of psychopathology. Importantly, it is essential to demonstrate the impact of these factors and symptoms through longitudinal, experimental and interventional research. Interventions targeting the reduction in fear of somatic sensations, attentional alterations toward bodily sensations, or self-examination might have a significant clinical impact as they might have the potential to alleviate other relevant symptoms and psychopathology in general. For instance, specialized intervention programs focused on alleviating anxiety sensitivity (i.e. cognitive anxiety sensitivity treatment [CAST; Schiele et al., 2021] and anxiety sensitivity amelioration training [ASAT; Schmidt et al., 2007]) or the elimination of anxiety maintaining safety and avoidance behaviors (False Safety Behavior Elimination Therapy [F-SET, Riccardi et al., 2017; Schmidt et al., 2012]) could be cost-efficient and brief treatments to prevent symptom escalation and achieve symptom reduction (i.e. in anxiety, depression, avoidance, and functional impairment).

Limitations

Several limitations of the current work should be discussed. First, only few subscales of questionnaires were used in the current study. For example, the physical concern subscale of the ASI-3 (Kemper et al., 2011) was selected to obtain an efficient network. The use of brief measures of depression and anxiety did not allow for a nuanced characterization of the interrelations among the full spectrum of depressive and anxiety symptoms as well as interoceptive fear and maladaptive attention and behaviors. Employing more comprehensive symptom assessments are recommended in future research to enhance our understanding of these complex interrelationships. With the present sample size, a network with more nodes would result in decreased specificity. However, due to the use of specific subscales, no further conclusions about other associations with symptoms or other relevant factors can be drawn. Second, the calculated network was based on cross-sectional data. Therefore, causal conclusions could not be drawn. Moreover, the present network analysis does not allow for the analysis of within-subject dynamics over time, thus allowing no differentiation between within- and between-subject variance (Bos et al., 2017; Epskamp, 2020; Hamaker, 2012). Future studies are warranted to prove the etiological relevance of the identified symptoms in the escalation of symptomatology. Third, this study analyzed a network based on data from a healthy cross-sectional sample. To draw further conclusions regarding the relative importance of symptoms for psychopathology, network centrality and bridge strength of symptoms in symptom networks of healthy and anxious individuals should be compared. Fourth, a high percentage of female participants completed the survey, which is common in voluntary survey research (Smith, 2008). Through further recruitment strategies a more diverse sample could be obtained in future research. Finally, our study was conducted during the COVID-19 pandemic, which might have influenced the present network and thus might limit generalization of the results to non-pandemic time points. However, it is important to note that our data collection occurred during a period of relative normalization of psychological distress in Germany, with levels of psychological burden comparable to both pre-pandemic and current times (Bendau et al., 2022; Mauz et al., 2023; Robinson et al., 2022). Notably, questionnaire scores in our study were comparable with those from pre-pandemic research, supporting the potential generalizability of our findings beyond the pandemic’s unique context. Moreover, as previous research has shown considerable inconsistencies in the identification of central and bridge symptoms due to the specificity of the sample (Jin et al., 2022), the present network should be validated with contemporary data. However, the absence of more contemporary data for network validation and the inclusion of pandemic-specific measures, such as fear of COVID-19, which may now be less relevant, present challenges for validating the current network. Despite these challenges, simulation studies and accuracy analyses (Epskamp et al., 2018) have confirmed the stability and reliability of our network. Future research should aim to replicate and extend our findings in various post-pandemic contexts to more thoroughly assess the network’s generalizability and stability.

Conclusion

In the present study, we identified fear of bodily sensations, time spent scanning the body for abnormalities, attention to anxiety-related sensations, and self-examination as key components within a comprehensive network associated with psychopathological symptomatology. These factors should be considered as important targets for future research which could inform future interventions and prevention strategies. This study suggests associations of maladaptive focus and the interpretation of somatic sensations with core symptoms of depression and anxiety. This should be further explored and verified in future studies, especially longitudinal studies. Investigation of temporal dynamics within a network can reveal further interrelations, especially in the context of psychotherapy, and thus, will contribute to the development of more effective, efficient, and personalized interventions in the future.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data and analysis code can be obtained from here (https://osf.io/5b29f/). Please contact the corresponding author (Gessner, J.) for further information.

Supplementary material

Supplemental material for this article can be accessed online at https://doi.org/10.1080/16506073.2024.2336036

Additional information

Funding

This work was supported by the DYNAMIC center, which is funded by the LOEWE program of the Hessian Ministry of Science and Arts [Grant Number: LOEWE1/16/519/03/09.001(0009)/98].