Anxiety and depression are among the most prevalent mental health issues in the general population currently, affecting millions of children and adolescents (World Health Organization, 2022). It has been estimated that the global prevalence of children experiencing clinically elevated depression and anxiety symptoms is around 20–25%, whereas the prevalence of any anxiety or depressive disorder is about 6% and 2%, respectively (Polanczyk et al., Reference Polanczyk, Salum, Sugaya, Caye and Rohde2015; Racine et al., Reference Racine, McArthur, Cooke, Eirich, Zhu and Madigan2021). The presence of high comorbidity in these conditions has been reported, especially in child and adolescent populations, noting that comorbidity tends to result in increased severity of symptoms, resistance to treatment, and increased risk for other problems (e.g., physical illnesses or suicide risk) (Melton et al., Reference Melton, Croarkin, Strawn and McClintock2016). Additionally, this comorbidity has been associated with other negative factors and consequences, such as stress, interpersonal difficulties, and interference with functioning in various areas (Shao et al., Reference Shao, He, Ling, Tan, Xu, Hou, Kong and Yang2020).
In this regard, there has been a growing interest in recent years in understanding various aspects of the comorbidity of anxiety and depression, such as shared factors (e.g., negative affectivity, interpersonal and cognitive processes) and developmental trajectories or pathways (Cummings et al., Reference Cummings, Caporino and Kendall2014). Despite a growing body of previous research that has revealed various possible correlations between the two conditions, the findings are mixed. Some findings support that the presence of anxiety symptoms is important in the long-term persistence of depressive symptoms, considering variables such as the number and severity of symptoms (Coryell et al., Reference Coryell, Fiedorowicz, Solomon, Leon, Rice and Keller2012). Essau et al. (Reference Essau, de la Torre-Luque, Lewinsohn and Rohde2020) found that factors like high levels of loneliness or state anxiety predicted a trajectory of increasing depressive symptoms, and this trajectory, in turn, predicted the development of anxiety disorders in adulthood. In this line of research, other authors have supported that both anxiety and depression in childhood or adolescence can predict subsequent anxiety or depression disorders, respectively (homotypic continuity), and have also found heterotypic continuity; that is, childhood anxiety predicted adolescent depression or depression predicted subsequent anxiety (Cohen et al., Reference Cohen, Andrews, Davis and Rudolph2018; Ranøyen et al., Reference Ranøyen, Lydersen, Larose, Weidle, Skokauskas, Thomsen, Wallander and Indredavik2018).
Hence, it is essential to enhance our understanding of the comorbidity between childhood anxiety and depression, involving key stakeholders (e.g., parents, teachers), and ultimately, striving to develop strategies for the early detection of implicated symptoms, as well as effective preventive and treatment interventions (Melton et al., Reference Melton, Croarkin, Strawn and McClintock2016). In this regard, a novel approach known as network analysis has emerged in recent years that may be valuable and effective for investigating and better understanding the relationship between symptoms of different conditions, including the comorbidity of anxiety and depression (McElroy et al., Reference McElroy, Fearon, Belsky, Fonagy and Patalay2018).
This methodology allows for estimating partial correlations between variables (nodes within the network), highlighting the unique variance between two variables after conditioning on all other variables (Epskamp & Fried, Reference Epskamp and Fried2018). One of the aspects it measures is centrality indices, with two indices, in particular, having gained more support for examining the importance of symptoms or nodes in the network. These include strength (i.e., summing the absolute edge weights of edges per node) and one-step expected influence (EI: Summing the edge weights connected to a node without using the absolute value) (Chung et al., Reference Chung, Yun, Nguyen, Rami, Piao, Li, Lee, Kim, Sui, Kim, Lee, Kim, Yu, Lee, Won, Lee, Kim, Kang and Kim2021; Fonseca-Pedrero et al., Reference Fonseca-Pedrero, Al-Halabí, Pérez-Albéniz and Debbané2022). Additionally, complementarily, it is also possible to identify symptoms that may play a significant role in connecting two different psychological issues or disorders, known as bridge symptoms, potentially contributing to the development and maintenance of comorbidity. Therefore, this network approach helps to better understand comorbidity between conditions through quantitative indicators (e.g., bridge strength or bridge EI), providing information about symptoms whose treatment or deactivation may prevent the spread of comorbidity and the subsequent negative consequences (Jones et al., Reference Jones, Ma and McNally2021). Furthermore, it is possible to visually identify the shortest pathways between nodes in a network and the mediating nodes that may be involved in the relationship between anxiety and depression symptoms (Isvoranu et al., Reference Isvoranu, van Borkulo, Boyette, Wigman, Vinkers, Borsboom and Investigators2017).
To date, various studies using network analysis with children and adolescents have yielded diverse results. Available studies tend to demonstrate the formation of highly interconnected networks, thus highlighting the strong association between anxiety and depression symptoms throughout development in both community and clinical populations (McElroy & Patalay, Reference McElroy and Patalay2019; McElroy et al., Reference McElroy, Fearon, Belsky, Fonagy and Patalay2018). There is considerable variability in the most central symptoms detected among studies, which may be partly attributed to using different measurement instruments (Fried, Reference Fried2017).
Authors have identified different central depressive and anxious symptoms in the network of anxiety and depression in children and adolescents. For instance, a study in a Spanish community sample revealed the significance of depressive symptoms such as “feels lonely,” “thinks life has been a failure,” and “feels sad,” as well as anxiety symptoms like “anxious,” “fears school” and “worries” (Sánchez-Hernández, Holgado-Tello, et al., Reference Sánchez-Hernández, Carrasco and Holgado-Tello2023). Another study in a Chinese community population found that central depressive symptoms were “sad mood” and “guilt,” whereas central anxious symptoms were “irritability” and “worry too much” (Cai et al., Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022). Mullarkey et al. (Reference Mullarkey, Marchetti and Beevers2019) emphasized central depressive symptoms like “self-hatred,” “loneliness,” “sadness,” and “pessimism” in a U.S. community sample. Authors like McElroy et al. (Reference McElroy, Fearon, Belsky, Fonagy and Patalay2018) found central anxiety-depression symptoms in a longitudinal study with the general U.S. population, including “fearful/anxious,” “sad/unhappy,” “nervous,” and “worthless,” which were generally stable over time.
In terms of specific analyses of bridge symptoms between childhood and adolescent anxiety and depression, studies also yield diverse results. For example, Cai et al. (Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022) found that all the bridge symptoms they identified, such as “guilt,” “sad mood,” and “suicide ideation,” belonged to the depressive symptom community. In contrast, the study by Sánchez-Hernández, Carrasco, et al. (Reference Sánchez-Hernández, Holgado-Tello and Carrasco2023) with a Spanish sample identified the primary bridge symptoms as “feeling lonely” and “feeling unloved.” Interestingly, these authors also estimated short pathways, showing the potential existence of routes for the connection between anxiety and depression, highlighting connections between fear-related symptoms and self-perception of worth and feeling unloved, or between symptoms related to interpersonal relationships and suicidal ideation.
The main goal of this study was to explore and understand the comorbidity between anxiety and depression symptoms in a sample of school-age children using a network analysis approach. Specifically, the objectives were to: (a) Examine the item-level relationships between symptoms of the two conditions by creating a comorbidity network and determining centrality indices to identify the most central nodes based on the estimation of Strength and EI centrality indices; (b) explore potential bridge symptoms in the network by calculating bridge centrality indices for each node, such as bridge strength or bridge EI, and creating a comorbidity network that graphically highlights the emerging bridge symptoms; (c) visually explore the shortest pathways between specific anxiety and depression symptoms in the sample by creating the shortest path network as an additional means to better understand the interconnection between the two conditions. Given the high comorbidity between anxiety-depression symptoms and based on previous research using a network analysis approach (e.g., Cai et al., Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022; McElroy et al., Reference McElroy, Fearon, Belsky, Fonagy and Patalay2018; Sánchez-Hernández, Carrasco, et al., Reference Sánchez-Hernández, Holgado-Tello and Carrasco2023; Sánchez-Hernández, Holgado-Tello, et al., Reference Sánchez-Hernández, Holgado-Tello and Carrasco2023), it was expected that: (a) A highly interconnected network of anxiety and depression symptoms would emerge, rather than two disconnected sets of nodes; (b) several nodes/symptoms belonging to both conditions (i.e., depression and anxiety) would emerge with greater centrality or relevance in the network; (c) various bridge symptoms would emerge, predominantly representing content related to cognitive, interpersonal, and/or negative affective states; and, (d) different possible short pathways would be found involving various types of anxiety and depression symptoms.
Method
Participants
The sample consisted of 281 children aged 8 to 12 years (M = 9.39, SD = 1.36), of whom 54.8% were boys and 45.2% were girls. The age distribution was as follows: 8 years (37%), 9 years (21%), 10 years (17.8%), 11 years (14.2%), and 12 years (10%). Most children (98.9%) were born in Ecuador, while only three were born in other Spanish-speaking countries (i.e., Colombia, Venezuela). Therefore, all study participants were Spanish speakers. The number of siblings per child ranged from none to five (M = 1.30, SD = 1.03), with the highest percentages having one sibling (43.1%), two siblings (23.5%), and no siblings (21.7%).
Regarding the relatives who served as informants in this study, the highest participation was from mothers (n = 242; 86.1%) and fathers (n = 28; 10%), followed by siblings (n = 8; 2.8%) and other family members (e.g., uncles, grandparents) (n = 3; 1.1%), who in these few cases, responded as the children’s primary caregivers. As for the age of the informants, the majority were in the age range of 35 to 44 years (46.3%), followed primarily by the age range of 25 to 34 years (36.3%) and 45 to 54 years (12.8%). Regarding the level of education, a higher level of education such as a bachelor’s degree (46.2%) or master’s/doctorate (27.8%) predominated, followed by secondary education (14.2%), basic education (10.7%), or being a student (1.1%). The marital status of the participants was predominantly married or living with a partner (71.4%), followed mainly by other statuses such as divorced or separated (13.9%) and single (13.9%). Lastly, the informants had various family income levels. Specifically, 12.9% of those who disclosed this information reported monthly net incomes equal to or exceeding €2,000, whereas the majority (58.3%) reported incomes below this threshold, suggesting a low-to-medium socioeconomic level.
Measures
The Short Mood and Feelings Questionnaire-Parent report (SMFQ-P; Angold et al., Reference Angold, Costello, Messer and Pickles1995). The 13-item brief version of the Mood and Feelings Questionnaire-Parent report (MFQ-P), commonly known as the SMFQ-P, is a validated tool completed by parents to assess the severity of depressive symptomatology in their children. Each item is formulated to address various manifestations (cognitive, emotional, and behavioral aspects) linked to childhood depression. For this purpose, parents are asked to rate the items on a three-point Likert scale (ranging from 0 = Not true to 2 = True), based on how their child felt or acted in the past two weeks, and the total score reflects the sum of all the items, with higher scores indicating higher severity. The SMFQ-P has shown good psychometric properties in other samples involving school-aged children, with reliability coefficients of .87 and .92 in the original study and the adaptation to Spanish-speaking population, respectively (Angold et al., Reference Angold, Costello, Messer and Pickles1995; Espada et al., Reference Espada, Belzunegui-Pastor, Morales and Orgilés2024). In the current sample, the internal consistency of the measure was excellent (Cronbach’s α = .90).
The Spence Children’s Anxiety Scale-Parent Report, brief version (SCAS–P–8; Reardon et al., Reference Reardon, Spence, Hesse, Shakir and Creswell2018). This abbreviated version of the original SCAS comprises eight items that encompass symptoms of several anxiety disorders, including separation anxiety, social anxiety, and panic/agoraphobia. Parents are asked to rate the items on a four-point Likert scale (ranging from 0 = Never to 3 = Always). A total score can be calculated, where higher scores indicate higher anxiety levels. The SCAS–P–8 has demonstrated good psychometric properties in other samples of school-aged children, with reliability coefficients of .82 in the original study and the adaptation to Spanish-speaking population (Orgilés et al., Reference Orgilés, Morales, Espada and Rodríguez-Menchón2022; Reardon et al., Reference Reardon, Spence, Hesse, Shakir and Creswell2018). In this sample, the measure demonstrated good internal consistency (Cronbach’s α = .82)
Procedure
An online survey was disseminated through social media platforms (i.e., Facebook, X, and Instagram) using the non-probabilistic snowball sampling method. Families were provided with study information and invited to participate following informed consent. Data was collected from a single informant (mother, father, or the children’s primary caregiver), and no incentives were provided for participation. Participation was anonymous. The survey required approximately 10 minutes to complete. The study obtained approval from the Ethics Committee of the authors’ institution (Ref. ADH.DES.MAIG.MAIG.23).
Data Analysis
To conduct the comorbidity network of anxiety and depression symptoms, we specified a Gaussian graphical model (GGM), an undirected network model of partial correlation coefficients, regularized using a graphical LASSO (GLASSO) algorithm in combination with the extended Bayesian information criterion (EBIC) model. Each item/symptom of the brief MFQ-P and SCAS-P represented a node within the network. The nodes represent the observed variables in this model, which is applied to psychological cross-sectional data. They are statistically connected with an edge if two variables are not independent after conditioning on all other variables. Blue edges indicate positive partial correlations, and red edges indicate negative partial correlations, while thicker and more saturated edges denote higher correlations (Epskamp & Fried, Reference Epskamp and Fried2018; Epskamp, Waldorp, et al., Reference Epskamp, Waldorp, Mõttus and Borsboom2018). We also investigated centrality indices to measure the importance of the nodes in the network in terms of direct connectivity. This study focused on strength and EI in keeping with prior studies (Chung et al., Reference Chung, Yun, Nguyen, Rami, Piao, Li, Lee, Kim, Sui, Kim, Lee, Kim, Yu, Lee, Won, Lee, Kim, Kang and Kim2021; Fonseca-Pedrero et al., Reference Fonseca-Pedrero, Al-Halabí, Pérez-Albéniz and Debbané2022).
It should be noted that EI is similar to strength but retains the negative and positive value of the edge weight. Thus, EI will differ from strength in the case of a network with negative edges, therefore helping to prevent distorting the interpretation. It is also more appropriate for measuring the strength of a node’s influence (Robinaugh et al., Reference Robinaugh, Millner and McNally2016). Additionally, we examined symptoms that may form bridges between the two domains/disorders assessed by computing bridge centralities of strength and EI —the sum of the positive or negative values of all the edges that connect a node to all the nodes that are not part of the same community- (Jones et al., Reference Jones, Ma and McNally2021). In keeping with prior research (Chung et al., Reference Chung, Yun, Nguyen, Rami, Piao, Li, Lee, Kim, Sui, Kim, Lee, Kim, Yu, Lee, Won, Lee, Kim, Kang and Kim2021), we determined bridge symptoms that play a primary role based on scores higher than the 80th percentile for the bridge EI metric (Chung et al., Reference Chung, Yun, Nguyen, Rami, Piao, Li, Lee, Kim, Sui, Kim, Lee, Kim, Yu, Lee, Won, Lee, Kim, Kang and Kim2021). The shortest pathways between symptoms of anxiety and depression symptoms were also computed (Isvoranu et al., Reference Isvoranu, van Borkulo, Boyette, Wigman, Vinkers, Borsboom and Investigators2017).
Finally, following Epskamp, Borsboom, et al. (Reference Epskamp, Borsboom and Fried2018), the robustness of the comorbidity network was examined by estimating the edge-weight accuracy and the stability of the centrality indices. To estimate the accuracy of the edge weights, we bootstrapped the 95% confidence intervals (CIs) around the edge weights using a non-parametric bootstrap with 1,000 iterations. In addition, to investigate the stability of the order of the centrality indices (strength and EI), we used the case-dropping subset bootstrap (1,000 iterations). To quantify the stability of the centrality indices, we computed the correlation stability coefficient (CS-coefficient). This coefficient quantifies the maximum proportion of cases that can be removed to retain, with a 95% certainty, a correlation of at least 0.7 between the original centrality indices found and the centrality of networks based on subsets of the data. As recommended, the CS-coefficients should not be below the cutoff of 0.25, and preferably above 0.5, to consider the indices stable after dropping one subset of the sample (Epskamp, Borsboom, et al., Reference Epskamp, Borsboom and Fried2018).
All data analyses were performed using R (version 4.2.3) to estimate and visualize the networks, bridges, centrality indices (reported as standardized values, z scores), shortest pathway, and edge-weight accuracy and stability of the network. The R-packages “bootnet,” “qgraph,” and “networktools” were used.
Results
The descriptive statistics for the items included are presented in Table 1. From these items, comorbidity networks of anxiety and depression symptoms were estimated, which will be presented below, with each item representing a node.
Table 1. Means, Standard Deviations, Skewness, and Kurtosis of the Anxiety and Depression Item scores
Note. M = Mean; SD = Standard deviation; SCAS = items/nodes belonging to the brief version of the Spence Children’s Anxiety Scale - Parent report; MFQ = items/nodes belonging to the short version of the Mood and Feelings Questionnaire - Parent report.
a Short label of each item/node used along the study.
Network Structure and Centrality Measures
In order to examine the relationships between anxiety and depression symptoms, a comorbidity network was first estimated, including the assessment of centrality indices such as strength and EI (Figure 1). Figure 1a shows how the comorbidity network visually suggests a clear interconnectedness among items from the two symptom communities. As for the relationships between anxiety and depression symptoms, the strongest significant connections were found between the nodes MFQ8–SCAS4 (edge weight = 0.24), MFQ12–SCAS7 (edge weight = 0.24), MFQ5–SCAS8 (edge weight = 0.17), and MFQ6–SCAS8 (edge weight = 0.16).
Figure 1. Anxiety-Depression Comorbidity Network and Centrality Indices.
Note. (a) Estimated anxiety-depression comorbidity network. Thicker solid lines (edges) represent higher connections between nodes. Blue lines represent positive regularized partial correlations; red lines indicate negative correlations; (b) centrality indices of strength and expected influence. Abbreviation: SCAS = nodes belonging to the brief version of the Spence Children’s Anxiety Scale - Parent report; MFQ = nodes belonging to the short version of the Mood and Feelings Questionnaire-Parent report.
In terms of EI (Figure 1b), it was observed that the most influential nodes in the network were, in the following order: MFQ5 (“No good anymore”; EI = 1.21), MFQ12 (“Could never be as good”; EI = 1.21), MFQ8 (“Hated self”; EI = 1.17), MFQ13 (“Did everything wrong”; EI = 1.16), MFQ11 (“Nobody loved him/her”; EI = 1.13), and SCAS8 (“Suddenly feels really scared”; EI = 1.05). The nodes SCAS 5 (“Trouble going to school”; EI = 0.95), SCAS7 (“Worries about what others think”; EI = 0.95), and SCAS2 (“Feels afraid”; EI = 0.91) also exhibited significant influence in the network, although not as high as the ones mentioned above. The centrality results in terms of strength were similar to those of EI, as shown in Figure 1b. The nodes related to depression symptoms were still the most important in the network, with slight differences in terms of anxiety symptoms, where the nodes SCAS2 and SCAS4 had higher values than in EI. However, this result may be distorted because these nodes have negative relationships, so focusing on the EI value may be more appropriate in this case.
Bridge Symptoms
Bridge EI was computed to identify symptoms bridging the two conditions or domains. The nodes that emerged as the most prominent bridge symptoms were nodes SCAS5 (“Trouble going to school”), SCAS8 (“Suddenly feels really scared”), MFQ12 (“Could never be as good”), and MFQ 8 (“Hated self”), defined as items scoring above the 80th percentile for the EI metric (Figure 2a). Although not as prominent as the previous ones, the bridge EI index indicated that node SCAS7 (“Worries about what others think”) may also be a relevant bridge symptom. The bridge strength index showed similar results, as reflected in Figure 2b, with the main notable difference being that node SCAS4 (“Afraid to make a fool of self”) stood out as the most prominent bridge symptom. However, this index sums absolute values and, therefore, does not consider negative values, which may explain the difference from the EI bridge result.
Figure 2. Bridge Symptoms in the Anxiety-Depression Comorbidity Network.
Note. (a) Estimated anxiety-depression comorbidity network showing bridge connections. Yellow nodes denote bridging symptoms that are key in linking the two domains, based on cutoff scores above the 80th percentile for the bridge expected influence (EI) metric. Abbreviations: SCAS = nodes belonging to the brief version of the Spence Children’s Anxiety Scale - Parent report; MFQ = nodes belonging to the short version of the Mood and Feelings Questionnaire - Parent report. Thicker solid lines (edges) represent stronger connections between nodes. Blue lines represent positive correlations; red lines indicate negative correlations; (b) centrality indices of bridge strength and EI.
Shortest Path Network
Regarding the estimated shortest path network (Figure 3), interesting and notable connections between symptoms were observed.
Figure 3. Shortest Path Network between Anxiety and Depression.
Note. Thicker solid lines indicate stronger connections. Blue and red lines represent positive and negative correlations, respectively. Dashed lines represent existing background links in the network that are less important in terms of shortest paths. SCAS = nodes belonging to the brief version of the Spence Children’s Anxiety Scale - Parent report; MFQ = nodes belonging to the short version of the Mood and Feelings Questionnaire - Parent report.
1. One of them is the circle found among SCAS7 (“Worries about what others think”), MFQ12 (“Could never be as good”), MFQ11 (“Nobody loved him/her”), MFQ10 (“Lonely”), MFQ8 (“Hated self”), SCAS4 (“Afraid to make a fool of self”), and back to SCAS7.
2. It was also observed how significant connections are formed between most anxiety symptoms and depression symptoms through the nodes SCAS5 and SCAS8. One of them is the circle composed of depressive symptoms like MFQ1 (“Miserable/unhappy”), MFQ2 (“Did not enjoy anything”), MFQ3 (“Tired”), MFQ6 (“cried a lot”), a circle that, through these last two nodes, also connects directly with anxiety symptoms through the nodes SCAS5 (“Trouble going to school”) and SCAS8 (“Suddenly feels really scared”), as well as with other depressive symptoms.
3. Similarly, remarkable connections were observed between MFQ11 (“Nobody loved him/her”), MFQ12 (“Could never be as good”), MFQ13 (“Did everything wrong”), and MFQ5 (“No good anymore”), depressive symptoms primarily related to content linked to negative thoughts and feelings at an interpersonal and self-esteem level. These were shown to be directly associated with anxiety symptoms (i.e., through connections from node MFQ12 to nodes SCAS7: “Worries about what others think” - social anxiety - or SCAS 5: “Trouble going to school” - separation anxiety - or node MFQ5 with SCAS8: “Suddenly feels really scared” - panic/agoraphobia). But they are also indirectly associated, as these anxiety symptoms connect with other anxiety symptoms. For example, the node SCAS7 connects directly with other anxiety symptoms (e.g., generalized anxiety symptoms measured by nodes SCAS1: “Worries about things,” SCAS2: “Feels afraid,” SCAS6: “Worries something bad will happen”), and SCAS5 connects directly with other anxiety symptoms (i.e., Node SCAS6 – generalized anxiety - or node SCAS8 – panic/agoraphobia).
Additionally, when examining the relationships between these depressive symptoms, it is worth noting the direct connection between MFQ5 and other depressive symptoms like MFQ9 (“Bad person”) strongly connected to MFQ8 (“Hated self”), which, in turn, connects with anxiety symptoms through the node SCAS4 (“Afraid to make a fool of self” – social anxiety).
Network Accuracy and Stability
Analyses suggested that the comorbidity network was accurately estimated, with moderate confidence intervals around the edge weights (Figure 4a). The stability of the strength and EI centrality indices was investigated through the case–dropping bootstrap analysis (Figure 4b). The CS–coefficients for the centrality indices of strength and EI were 0.52, thus exceeding the recommended threshold for a stable estimation of 0.5.
Figure 4. Network Accuracy and Stability.
Note. (a) Bootstrapped 95% confidence intervals (CIs) of estimated edge-weights for the estimated anxiety-depression comorbidity network. The red line indicates the sample values and the gray area represents the bootstrapped CIs. Each horizontal line represents one edge of the network, ordered from highest edge weight to lowest edge weight; (b) Case-dropping bootstrap for the network. The x-axis indicates the percentage of cases used in the analysis. The y-axis represents the average correlations between the estimated centrality indices (strength and expected influence) of the original network and the centrality indices from re-estimated networks after dropping increasing percentages of cases. The lines represent the correlations of the centrality indices.
Discussion
This study aimed to examine item–level relationships between anxiety and depression symptoms in a sample of school–age children using network analysis.
The comorbidity network model estimated indicated, as expected, a considerable interconnection between symptoms in both domains, highlighting relationships between “Hated self” (MFQ8) and “Afraid to make a fool of self” (SCAS4); “Could never be as good” (MFQ12) and “Worries about what others think” (SCAS7); and “Suddenly feels really scared” (SCAS8) with “No good anymore” (MFQ5) and “Cried a lot” (MFQ6). This shows that the symptoms that were strongly and directly connected were depressive symptoms—primarily reflecting cognitive and interpersonal aspects of depression—with symptoms of various anxiety problems, particularly panic/agoraphobia and social anxiety. This is consistent with previous literature demonstrating comorbidity between different specific anxiety problems or disorders and depression, emphasizing the relationship between social anxiety and depression (Cummings et al., Reference Cummings, Caporino and Kendall2014; de la Torre–Luque et al., Reference de la Torre-Luque, Fiol-Veny, Balle, Nelemans and Bornas2020).
Besides, as hypothesized, the central nodes included symptoms of depression and anxiety. Specifically, the analyses indicated that the nodes “No good anymore” (MFQ5) and “Could never be as good” (MFQ12) were the most central and influential in the estimated network model, followed by “Hated self” (MFQ8), “Did everything wrong” (MFQ13), “Nobody loved him/her” (MFQ11), and “Suddenly feels really scared” (SCAS8). The stability of the estimated centrality indices was good, showing that strength and EI were stable, with CS–coefficients higher than 0.5 (Epskamp, Borsboom, et al., Reference Epskamp, Borsboom and Fried2018). It is worth noting that the central symptoms were mainly depressive symptoms, suggesting that in the anxiety–depression relationship, depressive symptoms may have a greater influence, at least in this sample. This is consistent with studies showing how depressive symptomatology can play a significant role in the development of anxiety problems. For example, findings indicate that depression or a trajectory of increased symptoms predicts the later development of anxiety disorders (Essau et al., Reference Essau, de la Torre-Luque, Lewinsohn and Rohde2020; Ranøyen et al., Reference Ranøyen, Lydersen, Larose, Weidle, Skokauskas, Thomsen, Wallander and Indredavik2018). However, it is important to note that there were nodes related to anxiety symptoms that also showed considerable centrality values and should, therefore, be considered. These nodes were mainly related to panic/agoraphobia symptoms (SCAS8), followed to a lesser extent by others such as school–related separation anxiety symptoms (SCAS5: “Trouble going to school”) or generalized anxiety symptoms (SCAS2: “Feels afraid”) and social anxiety (SCAS7: “Worries about what others think”) as measured by the SCAS–P–8. This is also in line with the literature showing the importance of anxious symptomatology in understanding the development or maintenance of comorbidity with depression, for example, as a predictor in the subsequent development of depression or higher and more persistent depressive symptoms (Cohen et al., Reference Cohen, Andrews, Davis and Rudolph2018; Coryell et al., Reference Coryell, Fiedorowicz, Solomon, Leon, Rice and Keller2012; de la Torre-Luque et al., Reference de la Torre-Luque, Fiol-Veny, Balle, Nelemans and Bornas2020).
Additionally, previous studies focused on examining these emotional problems using network analysis have also found high centrality levels in depressive and anxiety symptoms related to school-related concerns, fears, or difficulties (Cai et al., Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022; Sánchez-Hernández, Holgado-Tello, et al., Reference Sánchez-Hernández, Carrasco and Holgado-Tello2023). More specifically, considering the most central depressive and anxious symptoms in our estimated network, the findings are particularly in line with studies that emphasized the centrality of symptoms such as “self-hatred,” “worthlessness,” or “fearful/anxious” (McElroy et al., Reference McElroy, Fearon, Belsky, Fonagy and Patalay2018; Mullarkey et al., Reference Mullarkey, Marchetti and Beevers2019). Although making direct comparisons between studies is challenging due to the use of different measurement instruments, it is noteworthy that the most central depressive symptoms in this study have slightly different content than other previous studies. For instance, in the current study, symptoms of depression, often found in previous research, such as those associated with feelings of sadness or loneliness as central features (Cai et al., Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022; McElroy et al., Reference McElroy, Fearon, Belsky, Fonagy and Patalay2018; Mullarkey et al., Reference Mullarkey, Marchetti and Beevers2019; Sánchez-Hernández, Holgado-Tello, et al., Reference Sánchez-Hernández, Holgado-Tello and Carrasco2023), did not emerge as the most central symptoms. Instead, the most central symptoms were linked to aspects like not feeling good/useful enough or not as good as other children, self-hatred, doing everything wrong, or not feeling loved enough by others. Although the explanation for these differences remains unclear, the discrepancy between studies could be partially due to the use of different measures of anxiety and depression between studies, leading to different research results (Fried, Reference Fried2017). Thus, our findings complement previous studies and suggest that symptoms not included among the core symptoms of depression, such as sadness or anhedonia (American Psychiatric Association, 2013), can also have a significant influence on the comorbidity between anxiety and depression and deserve attention. In this case, the most central depressive symptoms in the estimated network with this sample seem to be more related to social aspects and low self-esteem. In this regard, it has been found that factors like low self-esteem could be a frequent predictor of anxiety and depression disorders, whereas poor social adjustment or social skills were predictors of depression (Essau et al., Reference Essau, de la Torre-Luque, Lewinsohn and Rohde2020).
Regarding the bridge symptoms, as hypothesized, we found that the anxiety symptoms “Trouble going school” and “Suddenly feels really scared” (separation anxiety and panic/agoraphobia symptoms as measured by SCAS–P–8, respectively), followed by the depressive symptoms “Could never be as good” and “Hated self,” were the most central bridge symptoms in connecting anxiety and depression symptoms in the network estimated in this sample. The activation of bridge symptoms is expected to lead to the activation of symptoms from the other domain, so they may play a significant role in the comorbidity or interconnection between symptoms from different symptom communities (Chung et al., Reference Chung, Yun, Nguyen, Rami, Piao, Li, Lee, Kim, Sui, Kim, Lee, Kim, Yu, Lee, Won, Lee, Kim, Kang and Kim2021; Sánchez-Hernández, Carrasco, et al., Reference Sánchez-Hernández, Carrasco and Holgado-Tello2023), in this case, comorbidity between anxiety and depression symptoms. It should be noted that the results suggested that two symptoms related to social anxiety based on SCAS–P–8 (i.e., “Worries about what others think” and “Afraid to make a fool of self”) could also play a role as bridge symptoms, so they may help explain anxiety-depression comorbidity and should be considered. These results differ from other studies with children and adolescents that have identified other central bridge symptoms such as depression or non-specific anxiety-depression symptoms (e.g., “feels lonely,” “feels unloved,” “guilty,” “sad mood,” and “suicide ideation”) (Cai et al., Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022; Sánchez-Hernández, Carrasco, et al., Reference Sánchez-Hernández, Holgado-Tello and Carrasco2023). However, in the study by Sánchez-Hernández, Carrasco, et al. (Reference Sánchez-Hernández, Carrasco and Holgado-Tello2023) with a Spanish sample, they also found other symptoms involved in the bridge system, such as “fears school” (anxiety), which might align more closely with the findings in this study. Therefore, our study shows that there are other symptoms of both anxiety and depression that may be important in explaining comorbidity. As with the central symptoms, variations in the bridge symptoms identified between studies could be partially explained by the reliance on different measures of anxiety and depression. Previous literature examining bridging symptoms in anxiety-depression networks tends to reveal variability across study samples with different characteristics (e.g., age, cultures, contexts) (Cai et al., Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022; Kaiser et al., Reference Kaiser, Herzog, Voderholzer and Brakemeier2021; Sánchez-Hernández, Carrasco, et al., Reference Sánchez-Hernández, Holgado-Tello and Carrasco2023). Hence, we emphasize the need for future research to clarify those key symptoms that may be generalizable from symptoms specific to particular groups and which factors may explain those differences (Cai et al., Reference Cai, Bai, Liu, Chen, Qi, Liu, Cheung, Su, Lin, Tang, Jackson, Zhang and Xiang2022).
Furthermore, it is worth noting that, collectively, the bridge symptoms of depression and anxiety in the current study tend to reflect specific negative affectivity (e.g., feeling scared, hated self). However, they also exhibit cognitive and interpersonal content (e.g., the thought that they could never be as good as other kids, worrying about what others think), which is also evident when considering the most central symptoms mentioned above. In this regard, taken together, our results are consistent with previous findings that emphasized the relevance of both intra- and interpersonal symptoms (Sánchez-Hernández, Holgado-Tello, et al., Reference Sánchez-Hernández, Carrasco and Holgado-Tello2023). In line with these authors, this suggests the importance of addressing children’s cognitive and contextual or interpersonal aspects in interventions, as they can be significant factors in developing or maintaining these psychological problems.
Regarding the shortest path network, as expected, it showed a variety of possible paths connecting the two conditions, reflecting the complexity of the anxiety-depression comorbidity, in line with previous findings in Spanish-speaking populations (Sánchez-Hernández, Carrasco, et al., Reference Sánchez-Hernández, Holgado-Tello and Carrasco2023). Among the different connections found, the involvement of one or several of the bridge symptoms detected in this study (i.e., SCAS5, SCAS8, MFQ12, MFQ8, SCAS7, SCAS4) stands out, providing further evidence of their potential substantial mediating role. Specifically, the connections detected include: (a) Connections suggesting a pattern of interconnections between concerns and fears about self-image or social behavior (e.g., concerns about what others think, thinking that they could never be as good as other children, fear of making a fool of themselves), as well as children’s negative thoughts and feelings about oneself and the environment (e.g., thinking that nobody really loves them, feeling lonely, self-hatred); (b) connections between depressive symptoms, including some key depression symptoms (e.g., related to sadness or not enjoying anything), with anxiety symptoms through direct connections with panic/agoraphobia and separation anxiety symptoms; (c) a complex connection of various depressive symptoms (e.g., negative thoughts and feelings at the interpersonal level, feelings of worthlessness or guilt, negative self-concept or self-esteem) directly or indirectly related to symptoms related to different anxiety problems (e.g., social anxiety, separation anxiety, panic/agoraphobia, generalized anxiety). Overall, these results suggest that anxiety and depression symptoms can connect through different paths, primarily through the activation of specific symptoms (bridge symptoms in this study), forming a complex network of interconnections that can explain the high comorbidity between childhood depression and several anxiety problems. These findings are also consistent with theories proposing various pathways to explain the comorbidity between anxiety disorders and depression (Cummings et al., Reference Cummings, Caporino and Kendall2014).
This study has several limitations to consider. Firstly, the type of sampling method used, as it involves a lack of control over the participants who completed the survey through social media platforms. Secondly, this study has a small sample size and mainly involves participants from Ecuador. Therefore, the findings may not be generalizable to other populations. Thirdly, it relies solely on parent-report measures; therefore, it would be interesting to include self-report measures in the future. In this regard, it could be worthwhile to investigate any potential variances across studies in the central and bridging symptoms within comorbid anxiety-depression networks based on the informant (e.g., self-reports versus parent-reports), as this could impact the perception and reporting of childhood mental health symptoms (Poulain et al., Reference Poulain, Vogel, Meigen, Spielau, Hiemisch and Kiess2020). Lastly, in this study, caregivers other than parents participated by completing the parent version of the instruments used, representing only a minority of the sample. In future studies employing a similar sampling methodology, it might be interesting to consider adapting the measures for other potential relatives and to collect data on how much time these relatives spend with the children.
Despite these limitations, the present study contributes to the existing literature on the comorbidity of childhood anxiety and depression symptoms through an innovative approach such as network analysis. To our knowledge, there are no prior studies that have addressed the comorbidity of anxiety and depression from this perspective with Ecuadorian school-age children using brief scales like the SMFQ-P and SCAS–P–8. The findings from this study not only provide additional evidence of a strong connection between anxiety and depression symptoms but also shed light on how this connection operates by revealing potential pathways between both conditions through the estimated shortest path network. Furthermore, it identifies a set of specific symptoms that stand out for their importance and influence in connecting symptoms. Among these, the identified bridge symptoms are particularly clinically relevant, as early detection and intervention for these specific anxiety and depression symptoms can help prevent the development and maintenance of comorbidity. In this regard, this type of study can contribute to designing and developing transdiagnostic interventions of great value in current clinical practice. It underscores the presence of symptoms that could serve as appropriate targets for transdiagnostic interventions aimed at preventing or treating anxiety and depressive symptoms in Spanish-speaking children. In addition, the findings of the present study warrant future research in this area, as it is necessary to replicate this study in other samples of Spanish-speaking children of school age and from different backgrounds to better understand the comorbidity between childhood anxiety and depression.
Acknowledgement
The authors would like to thank all the participants in this study.
Authorship credit
I.F.M.: Data curation, methodology, formal analysis, writing – original draft; A.I.G.: Data curation, investigation, writing – review & editing; M.O.A.: Conceptualization, supervision, writing – review & editing.
Data Sharing
The data from this study are not publicly available.
Conflicts of Interest
None.
