Heterogeneity of depression and anxiety among college students and related risk and protective factors: a network model

doi:10.21203/rs.3.rs-5156973/v1

Download PDF

Article

Heterogeneity of depression and anxiety among college students and related risk and protective factors: a network model

https://doi.org/10.21203/rs.3.rs-5156973/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

Comorbidity of depression and anxiety is common among adolescents and can lead to adverse outcomes. However, there is limited understanding of the latent characteristics and mechanisms governing these disorders and their interactions. Moreover, few studies have examined the impacts of relevant risk and protective factors.

Methods

This cross-sectional study involved 1,719 students. Mplus 8.0 software was used to conduct latent profile analysis to explore the potential categories of depression and anxiety comorbidities. R4.3.2 software was used to explore the network of core depression and anxiety symptoms, bridge these disorders, and evaluate the effects of risk and protective factors.

Results

Three categories were established: “healthy” (57.8%), “mild depression-mild anxiety” (36.6%), and “moderately severe depression-moderate anxiety” (5.6%). “Depressed mood”, “nervousness”, and "difficulty relaxing" were core symptoms in both the depression-anxiety comorbidity network and the network of risk and protective factors. Stress perception and neuroticism serve as bridging nodes connecting some symptoms of depression and anxiety and are thus considered the most prominent risk factors.

Conclusions

According to the core and bridging symptoms identified in this study, targeted intervention and treatment can be provided to groups with comorbid depression and anxiety, thereby reducing the risk of these comorbidities in adolescents.

Biological sciences/Psychology

Health sciences/Health care

Health sciences/Risk factors

Depression

Anxiety

Latent profile analysis

Network analysis

Depression and anxiety have long been widespread mental health problems among adolescents. An annual review of studies on the prevalence of mental disorders in children and adolescents globally revealed that the prevalence of anxiety disorders was 6.5%, and the prevalence of depressive disorders was 2.6% [1]. The Systematic Global Burden of Disease Study conducted in 2019 showed that both depression and anxiety ranked among the top ten causes of burden-of-disease injuries in adolescents aged 10–24 years [2]. Thus, it is urgent and necessary to pay more attention to adolescents’ mental health. Studies have also shown that depression and anxiety lead to various adverse outcomes, such as interpersonal distress, impaired emotional perception and regulation, and academic and employment limitations [3, 4]. In addition, depression and anxiety are highly comorbid [5, 6], and those with these comorbidities have symptoms of both depression and anxiety, often causing more harm than depression or anxiety alone. This can lead to more frequent anxiety attacks, more severe symptoms, and even an increased risk of suicide [7, 8]. Overall, the issue of depression and anxiety among teenagers is deeply concerning and requires further investigation. Price et al. found that a treatment regimen targeting both diagnoses may exhibit better efficacy than a treatment targeting only one of these disorders [5]. Therefore, it is valuable to further elucidate the mechanism governing the comorbidities of depression and anxiety to explore ways to identify them and effectively intervene in vulnerable populations.

Anxiety and depression are heterogeneous [9], meaning that individuals with similar diagnoses may have different symptoms. However, most studies have applied a variable-centered approach, which treats samples as homogeneous [10], ignores the inherent heterogeneity of individuals, and can interfere with the analysis of different subgroups of depression and anxiety comorbidities. This may hinder targeted identification and intervention. The present study adopts an individual-centered approach, which classifies individuals into subgroups based on their responses to each extraneous variable, thereby accounting for group heterogeneity [11]. The research methods commonly used in the individual-centered approach include mean segmentation, latent profile analysis (LPA), cluster analysis, and latent profile modeling. Recently, many researchers have used LPA to identify potential categories of depression and anxiety. For example, Lei et al. detected four categories in a sample of Japanese adults: “depression and anxiety”, “depression and moderate anxiety”, “mild anxiety”, and “ most healthy” [12]. Sun et al. proposed a three-profile model comprising a “healthy group”, “anxiety group”, and “depression-anxiety group” [13]. Latent profile analysis is an emerging statistical method for continuous variables that assigns individuals to distinct categories based on specific probabilities. Depression and anxiety are significant predictors of adolescent mental health [14], and understanding their heterogeneity may facilitate early identification and intervention to help young populations deal with mental health problems early and promptly, thereby preventing further worsening of various important but easily overlooked symptoms.

Recently, networks have begun to emerge in psychology, psychiatry and other fields. This allows us to visualize comorbidities as a network of symptoms, thus providing a new perspective [15]. During diagnosis, mental disorder comorbidities are often considered separate and distinct disorders with no direct link between symptoms [16]. From the network perspective, mental disorders can be viewed as a network comprising interactions between the symptoms; specifically, symptoms are the nodes, and symptom-to-symptom interactions are the edges [15]. Network analysis evaluates the role of nodes in the entire network and their impacts on other nodes via indicators describing node characteristics. One study statistically evaluated a network model of depression and anxiety symptoms and classified the core symptoms as “depressed mood”, “worry out of control”, and “overly worried”, and the bridge symptoms as “depressed mood”, “restless”, and “psychomotor disorder” [17]. Similarly, Garabiles et al. [18] identified three bridge symptoms of “fatigue”, “depressed mood”, and “anhedonia”, and three core symptoms of “fatigue”, “overly worried”, and “depressed mood” while exploring comorbidities between anxiety and depression in Filipino migrant domestic workers. In these correlational networks of depression and anxiety, the indicators (i.e., core and bridging symptoms) represent potential targets for intervention, thus providing insights regarding the interactions and connections between symptoms [19–21]. Therefore, network analysis can effectively supplement category models to describe individual psychological traits. This can facilitate the identification of core and bridging symptoms for comorbidities to elucidate the mechanisms underlying the comorbidity between these disorders. This would further allow the development of more effective intervention measures or the proposal of new therapies to prevent and treat these disorders.

Numerous studies exploring the causes of depression and anxiety have indicated that both internal characteristics and external factors can trigger changes in mental health levels. Therefore, it is necessary to understand the mechanism whereby relevant risk or protective factors influence depression and anxiety within the symptom network. However, most studies have focused on how one or a few risk factors impact depression and anxiety, while research on protective factors also lacks comprehensive assessments of multiple factors. Indeed, relevant studies have provided insights regarding the effects of social support [22], prosocial behavior [23],executive function [13], emotional regulation [19], COVID-19 [24], and other factors on depression and anxiety. The present study focuses on three risk factors (stress perception, neuroticism, and interpersonal distress) and two protective factors (general self-efficacy and perceived social support). Based on a thorough review of the relevant literature, we apply network analytical methods to study the interactions between multivariate variables and depression and anxiety symptoms. This study evaluates multiple protective factors and risk factors to explore their relationships with depression and anxiety in order to improve the accuracy of individual mental health status assessments.

This study aimed to explore various categories of depression and anxiety by analyzing participant profiles according to their questionnaire scores to distinguish depression and anxiety comorbidity risk groups. Herein, we also investigate the core and bridging symptoms of depression and anxiety using a network model. Finally, we examine the network relationships between depression and anxiety and the established risk and protective factors (i.e., stress perception, neuroticism, interpersonal distress, perceived social support, and general self-efficacy).

Participants

Convenience sampling was used to obtain data from a questionnaire survey of students at a higher vocational college in Hebei Province, China. A total of 1,824 questionnaires were distributed, of which 1,719 were valid, corresponding to, an effective rate of 94.24%. Questionnaires were excluded as invalid if they were incomplete, if the checkbox options were all the same, or if the completion time was less than 300 seconds. The average age of the valid respondents was 18.23 ± 0.96 years; 559 were male; all were first-year students. The participants were informed of the survey purpose, confidentiality principles, and other related matters before providing their consent.

Measurements

Depression and Anxiety

Participants completed the 9-item Patient Health Questionnaire (PHQ-9) to self-assess their degree of depression. The items were scored on a 4-point Likert scale, with 0–4 being no depression, 5–9 being mild depression, 10–14 being moderate depression, 15–19 being moderately severe depression, and 20–27 being severe depression. The internal consistency and reliability were evaluated based on Cronbach's alpha coefficient for the scale, which was 0.863 [25].

The 7-item Generalized Anxiety Disorder (GAD-7) scale was used to assess each participant’s anxiety level. The items were scored on a 4-point Likert scale, with 0–4 being no clinically significant anxiety, 5–9 being mild anxiety, 10–14 being moderate anxiety, and ≥ 15 being severe anxiety. The internal consistency and reliability were evaluated based on Cronbach's alpha coefficient for the scale, which was 0.903 [26].

Risk and Protective Factors

This study used the Perceived Stress Scale (PSS-10), the Big Five Inventory-2 (BFI-2)—Neuroticism subscale, and the Interpersonal Comprehensive Diagnostic Scale (ICDS) to measure the three risk factors of depression and anxiety, i.e., stress perception, neuroticism, and interpersonal distress. Additionally, the General Self-Efficacy Scale (GSES) and the Perceived Social Support Scale (PSSS) were used to measure the two protective factors for depression and anxiety, i.e., general self-efficacy and perceived social support.

The PSS-10, developed by Cohen et al. [27], measures an individual's perception of stress in terms of their stress tolerance and other related characteristics. This scale contains 10 items scored on a 5-point Likert scale, with questions 4, 5, 7, and 8 being reverse-scored. Otherwise, higher scores indicate more pronounced psychological stress. The internal consistency and reliability were evaluated based on Cronbach's alpha coefficient for the scale, which was 0.816.

The BFI-2 includes 60 entries and five dimensions (extroversion, agreeableness, openness, neuroticism, and conscientiousness). The Neuroticism subscale has 12 questions scored on a 5-point Likert scale. Entries 4, 9, 24, 29, 34, 39, 44, and 49 are reverse-scored, but otherwise, higher scores indicate a higher level of neuroticism. The internal consistency and reliability of the scale were evaluated based on Cronbach's alpha coefficient, which was 0.876 [28].

The ICDS was developed by Zheng et al. [29] and used to assess individuals' distress during interpersonal interactions on a 28-item scale that includes four dimensions (interpersonal conversation, interpersonal communication, dealing with people, and dissimilarity). A 2-point scale is used, where a higher total score indicates a higher level of interpersonal distress. The internal consistency and reliability of the scale were evaluated based on Cronbach's alpha coefficient, which was 0.899.

The GSES was initially compiled by Schwarzer et al. [30] and later translated and revised by Wang et al. [31] to assess individuals' general self-efficacy. The scale contains 10 questions scored using a 4-level Likert method. The total score is calculated by adding the scores of the 10 questions and dividing by 10. The higher the score, the greater the individual's general self-efficacy. The internal consistency and reliability of the scale were evaluated based on Cronbach's α coefficient, which was 0.883.

The PSSS was compiled by Blumenthal et al. [32] and reflects the degree of social support individuals feel. The scale includes 12 questions encompassing three sub-dimensions (family, friends, and other support) scored using a 7-level Likert method. The higher the total score, the higher the level of perceived social support. The internal consistency and reliability of this scale were evaluated based on Cronbach's α coefficient, which was 0.939.

Statistical Analysis

Common method bias tests, descriptive statistical analysis, correlation analysis, and multivariate logistic regression were performed using SPSS 23.0 software.

Mplus 8.0 software was used to perform LPA of depression-anxiety comorbidities to determine the latent categories and distribution. We started with one category and gradually increased the number of categories for LPA until we attained the optimal model. Six metrics (i.e., AIC, BIC, aBIC, LMR, BLRT, and Entropy) were combined to determine the best-fitting model for the potential classes [33–36]. The values of AIC, BIC, and aBIC were used for model comparisons, where the smaller the value, the better the model fit. The values of LMR and BLRT reached the significance level of p < 0.05, which indicated that adding one more profile significantly improved the model fit. Entropy reflected the classification accuracy of the model, and Entropy ≥ 0.8 suggested that the classification accuracy was greater than 90% [37].

Multiple network model construction and computations were carried out using R 4.3.2 software employing a partial correlation coefficient based on the introduction of a penalty factor, i.e., the Image Least Absolute Value Convergence and Selection Algorithm (GLASSO) [38] to increase specificity and efficiency [39]. Green lines connecting nodes indicate positive correlations, whereas red lines indicate negative correlations. Edge weights are denoted by the thickness of the line, such that more correlated edges are thicker. Next, the four centrality metrics of the GLASSO network were calculated to evaluate the importance of each node in the network [40]. We also calculated the Bridge Expected Impact (BEI) to identify the bridge symptoms in the network [41].

Correlation stability (CS) coefficients were calculated using the bootnet package to assess the stability of the centrality and bridging centrality indices. The CS coefficients were computed by running a 727-case discard subset bootstrap indicating what proportion of participants could be discarded from the analysis to ensure that the correlation between the original and new centrality indices would be no less than 0.7 [42].

Common Method Bias Test

Considering that all of the data for this study came from the questionnaires, Harman's one-factor method was used to test for possible common method bias. The results showed that 15 factors had eigenvalues greater than 1. The first factor explained 24.6% of the variance, which is less than the 40% threshold proposed by Podsakoff et al. [43]. It is therefore reasonable to infer that the study did not have significant common method bias and meets the statistical requirements.

Descriptive Statistics

Table 1 presents the descriptive statistics and correlation coefficients for stress perception, neuroticism, general self-efficacy, perceived social support, interpersonal distress, and anxiety and depression. The results showed that perceived social support and general self-efficacy were significantly and negatively correlated with depression and anxiety, whereas interpersonal distress, neuroticism, and stress perception were significantly and positively correlated with depression and anxiety.

Table 1

Descriptive statistics and correlation analysis of each variable
Variable	M	SD	1	2	3	4	5	6	7
1 = Stress Perception	25.040	6.007	1
2 = Neuroticism	34.740	7.614	0.688***	1
3 = General Self-Efficacy	2.666	0.407	-0.411***	-0.493***	1
4 = Perceived Social support	62.150	12.348	-0.389***	-0.398***	0.296***	1
5 = Interpersonal Distress	8.890	6.064	0.550***	0.582***	-0.430***	-0.400***	1
6 = Anxiety	3.670	3.823	0.627***	0.618***	-0.336***	-0.366***	0.588***	1
7 = Depression	5.130	4.311	0.618***	0.585***	-0.346***	-0.410***	0.573***	0.808***	1
***P < 0.001

Latent Profiles of Depression and Anxiety Symptoms

Model Optimization

In this study, five potential profile models were selected for fitting analysis, and the fitting indices are shown in Table 2. The values of AIC, BIC, and aBIC decreased as the number of classes increased and were lower and similar when they were divided into 3, 4, and 5 categories. The smaller the AIC, BIC, and aBIC values, the better the model fits. In contrast, the larger the Entropy value, the better the model fit, and in general, the closer the Entropy value is to 1, the more accurate the classification. LMR and BLRT were significant, indicating that the addition of a profile significantly improved the model fit. The Entropy value was the highest when divided into 5 categories, followed by 3 categories and finally 4 categories, and the LMR and BLRT values of several groups were significant. However, the probabilities corresponding to 4 and 5 categories were less than 5%, and the proportion of categories needed to be higher, which was of no practical significance. On the basis of these simulations, the models were divided into three optimal types. Table 3 shows that the average attribution probability for different potential categories is between 96%-99%, indicating that the results of these three types of models are credible.

Table 2

Latent profile models for depression and anxiety
Model	AIC	BIC	aBIC	LMR	BLRT	Entropy	Probability
1	57035.861	57210.245	57108.585
2	48215.842	48482.868	48327.200	0.0000	0.0000	0.938	0.693/0.307
3	44873.127	45232.794	45023.119	0.0000	0.0000	0.942	0.578/0.366/0.056
4	43712.783	44165.092	43901.410	0.0334	0.0000	0.902	0.442/0.302/0.212/0.044
5	42162.650	42707.600	42389.911	0.0017	0.0000	0.957	0.517/0.043/0.204/0.198/0.038

Table 3

Average attributions for each latent class
class	1	2	3
1	0.978	0.022	0.000
2	0.037	0.960	0.003
3	0.000	0.008	0.992

Naming Latent Classes

The scores of the three potential classes in terms of the 16 dimensions of depression and anxiety are shown in Fig. 1. The average scores of depression and anxiety in class 1 were 2.44 and 1.12, respectively, and this class exhibited almost no symptoms of depression or anxiety. The class 2 scores in each dimension were significantly higher than those of class 1. Class 2 comprised 36.6% of the participants, whose average scores of depression and anxiety were 7.86 and 6.16, respectively. The scores of class 3 in each dimension were significantly higher than those of the other two categories. Class 3 comprised 5.6% of the participants, and the average scores of depression and anxiety were 15.53 and 13.8, respectively. According the PHQ-9 and GAD-7 scoring criteria (Fig. 1), class 1, 2, and 3 were named “healthy”, “mild depression-mild anxiety”, and “moderately severe depression-moderate anxiety”, respectively. In this study, class 2 and class 3 were considered risk groups for depression and anxiety comorbidity.

Network Analysis

Symptoms of Depression and Anxiety

A network analysis was performed for the risk groups for depression and anxiety comorbidities. The weights of the edges and the positive and negative correlations between the nodes are illustrated in Fig. 2, where GAD2 (“worry out of control”) and GAD3 (“overly worried”) are most closely connected to anxiety items, and PHQ4 (“fatigue”) is most closely connected with PHQ1 (“anhedonia”) among depression items. The strongest connection is between GAD1 (“nervousness”) and PHQ2 (“depressed mood”). Overall, depressive symptoms and anxiety symptoms are closely related, and each symptom affects at least one other.

The centralization results (Fig. 3a) indicate that PHQ2 (“depressed mood”) had the highest EI value among the depression items and had high closeness and strength, highlighting it as a core symptom of depression; it also had the most significant impact on each node in the depression network. Among the anxiety items, GAD4 (“difficulty relaxing”) has the highest EI value, followed by GAD1 (“nervousness”), which also had higher closeness and strength. Similarly, these can be regarded as the core symptoms of anxiety because they had the most significant impacts on each node in the anxiety network. Figure 3b shows that PHQ2 (“depressed mood”) had the highest BEI value and was the most prominent bridging symptom among the depression items. Figure 2 highlights the strong relationship between PHQ2 (“depressed mood”) and many symptoms of anxiety, including GAD1 (“nervousness”), GAD3 (“overly worried”), GAD4 (“difficulty relaxing”), and GAD6 (“irritable”), followed by PHQ9 (“suicidal ideation”), which plays a vital role in the comorbid pathogenesis of depression and anxiety. The most prominent bridging symptom of anxiety was GAD4 (“difficulty relaxing”). Figure 2 indicates that GAD4 (“difficulty relaxing”) had high weights corresponding to its edges with PHQ2 (“depressed mood”), PHQ3 (“sleep disturbance”), PHQ4 (“fatigue”), and PHQ9 (“suicidal ideation”), followed by GAD5 (“restless”). Indeed, most of the depression and anxiety nodes had high BEI values, meaning they can be considered as bridging symptoms that support the comorbidity network of depression and anxiety. According to the stability test, the centrality index of each node and the CS coefficient of the BEI are both 0.7, indicating that after 70% of the data are discarded, the index is still related to the original data (r = 0.7); thus, the network stability is high.

Risk and Protective Factors

Based on the network of depression and anxiety comorbidity, the incorporation of risk and protective factors for network analysis (Fig. 4) revealed that all factors were connected, with neuroticism and stress perception having the strongest correlations with depression and anxiety. Notably, there was a strong positive correlation between stress perception and neuroticism. GAD1 (“nervousness”) and GAD4 (“difficulty relaxing”) represented the EI maxima among anxiety symptoms, whereas PHQ2 (“depressed mood”) exhibited the EI maximum among depression symptoms (Fig. 5a); these factors had the strongest impacts on their respective communities, consistent with the results from the network analysis of depression and anxiety symptoms. Stress perception had the highest EI value among the factors influencing depression and anxiety comorbidity. Stress perception also has the highest BEI value, suggesting that it had the strongest bridging effect (Fig. 5b). Overall, it had a substantial positive correlation with depression and anxiety, followed by neuroticism, which also played a strong bridging role. Meanwhile, interpersonal distress also played a bridging role, although both its bridging role and correlation with depression and anxiety were weaker. The BEI values of perceived social support and general self-efficacy were both extremely low and negative, suggesting that their associations with other nodes were negatively correlated, and there was a positive correlation between the two.

The network stability was very high, with both the node and BEI CS coefficients equal to 0.7; this result indicates that after discarding 70% of the data, both metrics remain relevant to the original data set (r = 0.70). The narrow 95% confidence interval for the edges supports the conclusion that the network exhibits high accuracy.

In this study, LPA of students from a vocational college in Hebei Province, China, indicated that their depression and anxiety symptom scores had prominent categorical characteristics, allowing them to be divided into three categories: “healthy” (class 1), “mild depression-mild anxiety” (class 2), and “moderately severe depression-moderate anxiety” (class 3). The fitting indices showed that the models fit well, further indicating that there were differences among these categories in terms of the depression and anxiety symptoms in college students. These results reflect the heterogeneity of depression and anxiety in these representative groups. Participants in the “healthy” group had few symptoms of depression and anxiety, consistent with previous studies [44]. Importantly, the present study did not identify a single depression or anxiety class. We inferred that depression and anxiety generally occur together, which is in line with the findings of Choi et al. [7] and Yu et al. [6]. One study explained this result in terms of cultural context; specifically, Asians may be afraid of being labeled as "weak in character", so they tend to emphasize somatic and anxiety symptoms while reporting depressed mood [45].

The network analysis of depression and anxiety symptoms revealed strong connections between the various symptoms, yet not all symptoms had equal importance. The core and bridging symptoms of depression and anxiety comorbidity include PHQ2 (“depressed mood”), GAD4 (“difficulty relaxing”), and GAD1 (“nervousness”). These results are consistent with previous reports [46–48], which proposed that core symptoms play a crucial role in maintaining the symptom network; as a result, targeting these core symptoms for intervention and treatment can help restore mental health levels. University students face complex issues involving academics, employment, and interpersonal and romantic relationships. As society continues to advance, the pressures to address these issues can easily lead to the emergence of depression and anxiety symptoms. In particular, “difficulty relaxing” signifies mental tension, “nervousness” represents anxiety about impending disasters, and “depressed mood” indicates a low psychological state. The high centrality of the impact of high expectations suggests that reducing these symptoms could disrupt the depression and anxiety symptom network, leading to the highest potential benefits in terms of alleviating anxiety and depression [49]. Specific measures can be implemented to alleviate the three core symptoms: “difficulty relaxing” can be relieved by visiting green parks [50] and adopting intelligent stress relief systems [51]; “nervousness” can be managed through restorative exercise [52], relaxation breaks [53], or cryotherapy (i.e., applying ice packs to the back of the neck to reduce tension) [54]; and “depressed mood” can be addressed through emotional therapy [55] and therapeutic alliances [56].

The network of risk and protective factors with anxiety and depression comorbidities showed that stress perception and neuroticism were the risk factors with the highest bridging performances. For example, the "nervousness" symptoms of generalized anxiety disorder and the "depressed mood" symptoms of depressive disorders are directly related to the first step (refer to Fig. 2); however, they are related through stress perception in the second step (refer to Fig. 4). On this basis, stress perception can be considered a common cause of depression and anxiety. It is also reasonable to conclude that stress perception is reinforced by an association between these two symptoms [57]. This concept is consistent with previous findings showing that neuroticism and stress perception are strongly and positively correlated with depression and anxiety and thought to be the bridge supporting depression and anxiety comorbidity [52, 58, 59].

In addition to facilitating associations between symptoms, a risk factor can influence comorbidities between two disorders because it has several associations with multiple symptoms of both mental health problems [60]. In the present study, stress perception was shown to be strongly associated with the “fatigue” and “self-abasement” symptoms of depression, as well as with the “difficulty relaxing” symptom of anxiety. These relationships may be interpreted as stress perception leading to “fatigue”, “self-abasement”, and “difficulty relaxing”, thus highlighting the direct link between these symptoms and the comorbid disorders. Alternatively, it is equally reasonable to have a stronger stress perception due to more severe symptoms of depression and anxiety. Previous studies have shown that depression and anxiety symptoms may contribute to stress perception problems and vice versa; however, there is limited knowledge regarding which specific symptom(s) play the central role in this phenomenon. Neuroticism showed similar results to stress perception. The “irritable” symptom of anxiety and the “depressed mood” and “sleep disturbance” symptoms of depression are related through neuroticism and have a high positive correlation with the “irritable” symptom. This result is consistent with a previous report [61] and may be interpreted as high levels of neuroticism may cause a person to experience depressed mood and sleep disturbances, making them understandably irritable. It is worth exploring the positive correlation between neuroticism and stress perception and how the two may interact with one another to affect the comorbid network of anxiety and depression. Previous studies have shown that neuroticism is associated with higher perceived stress, regardless of whether COVID-19-related stressors are present [62]. Meanwhile, the network diagram of risk and protective factors and anxiety and depression indicated that interpersonal distress is another risk factor that plays a bridging role. However, the impact is not significant and relies on other factors [63].

The results of this study also indicate that perceived social support and general self-efficacy are protective factors of depression and anxiety. In other words, these factors are inversely associated with some symptoms of depression and anxiety. These findings suggest that perceiving social support or having a strong sense of self-efficacy may reduce the severity of symptoms. Notably, support from family and loved ones appears to be the key to addressing negative psychopathological symptoms [57, 58]. Perceived social support was positively correlated with general self-efficacy, which may directly enhance its protective effect. This provides insights regarding the potential utility of using well-being traits to detect and prevent mental health issues. The link between social support and mental health problems, as well as the role of social support as a key protective factor in depression and anxiety, has been documented previously [57, 64, 65].

The findings of this study also have several clinical implications. The "depressed mood", "difficulty relaxing", and "nervousness" symptoms are the most important because they contribute to the development and continuation of comorbid depression and anxiety, as demonstrated in previous studies [66–68]. Therefore, targeting these core symptoms, thereby inhibiting interactions between disorders may be a valuable strategy for treating depression and anxiety comorbidities. Similarly, the results suggest that stress perception problems and neuroticism can lead to the simultaneous occurrence of depression and anxiety. Therefore, experiencing these factors may increase the risk of depression and anxiety comorbidities in adolescents. Early intervention for adolescents to prevent stress perception problems and high risk of neuroticism may reduce the incidence of such comorbidities [60, 63].

In summary, this study explored the heterogeneity of depression and anxiety through LPA and constructed the "depression-anxiety symptom network" and the "influencing factors related to comorbid depression and anxiety" in the non-healthy groups. We studied the core symptoms, bridging symptoms, and factors influencing comorbidity of depression and anxiety. The results provide support for future research and guidance for interventions to treat depression and anxiety in adolescents. This study also offers a novel perspective on prevention and intervention in adolescents who are at risk of depression and anxiety or have already developed depression and anxiety comorbidities. The core and bridging symptoms identified in this study are valuable targets for formulating new treatment options and improving available interventions. Our inclusion of numerous influencing factors of depression and anxiety comorbidities expands the scope of previous studies, which can help to better understand these comorbidities to reduce their occurrence.

Inevitably, there are some limitations to this study. First, this study assessed students at one specific university, so the results should be interpreted and generalized (e.g., to the entire adolescent population) with caution. The symptom network structure may be affected by different sample populations, and therefore, future studies should increase the sample size and variability to obtain a more comprehensive and stable network structure, which could provide even more reliable conclusions. Second, the data used in this study were cross-sectional. They cannot determine the directionality of the margins, and therefore, it is impossible to determine causal relationships between symptoms. Similarly, it cannot reveal dynamic changes between individual symptoms, so longitudinal data is needed for future studies. Third, the network structure is specific to the questionnaires used in this study, which means that different assessment questionnaires may result in different network structures. Moreover, the network analysis was affected by the number of included nodes and calculation methods, and thus, the results should be interpreted cautiously before implementation in clinical practice. Finally, because depression and anxiety symptoms were assessed according to self-reported scales, the effect of recall bias on the outcomes could not be avoided. Therefore, future studies could use more objective indicators or combine various types of reporting questionnaires to provide more comprehensive and objective information.

LPA: Latent profile analysis

BIC: Bayesian information criterion

aBIC: adjusted BIC

AIC: Akaike’s information criterion

BLRT: Bootstrap likelihood ratio test

LMR: Lo-Mendell-Rubin

EI: Expected influence

CS: Correlation stability

BEI: Bridge expected influence

Acknowledgements

Sincere thanks to all the participants who completed the questionnaire.

Authors' contributions

C.J.N.: Funding acquisition, Project administration, Writing-review & editing; Y.Y.J.: Conceptualization, Formal analysis, Writing-original draft; Y.H.L: Conceptualization, Writing-original draft; X.D.W: Visualization, Methodology, Software; Y.P.: Investigation, Supervision; H.Y.Z., Z.S.C., X.R.L. & X.Z.: Writing-review & editing. All authors had reviewed the manuscript.

Funding

Humanities and Social Science Research Project of Hebei Education Department (SQ2023251).

Availability of data and materials

The data from this study is available from the corresponding authors upon reasonable request.

Ethics approval and consent to participate

Informed consent was obtained from all participants and the study procedures were carried out in accordance with the Declaration of Helsinki guidelines. Ethical approval was obtained from the Ethics Committee of China University Of Science And Technology (N0.2021037).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Polanczyk, G. V., Salum, G. A., Sugaya, L. S., Caye, A. & Rohde, L. A. Annual research review: A meta-analysis of the worldwide prevalence of mental disorders in children and adolescents. J. Child. Psychol. Psychiatry. 56, 345–365. 10.1111/jcpp.12381 (2015).
Global burden of 369. diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 396, 1204–1222, doi: (2020). 10.1016/s0140-6736(20)30925-9
Morales-Muñoz, I. et al. Impact of anxiety and depression across childhood and adolescence on adverse outcomes in young adulthood: A UK birth cohort study. Br. J. Psychiatry. 222, 212–220 (2023).
van Tol, M. J. et al. Regional brain volume in depression and anxiety disorders. Arch. Gen. Psychiatry. 67, 1002–1011 (2010).
Price, M., Legrand, A. C., Brier, Z. M. & Hébert-Dufresne, L. The symptoms at the center: Examining the comorbidity of posttraumatic stress disorder, generalized anxiety disorder, and depression with network analysis. J. Psychiatr. Res. 109, 52–58 (2019).
Yu, P. et al. Variety ACEs and risk of developing anxiety, depression, or anxiety-depression co-morbidity: the 2006–2022 UK Biobank data. Front. Psychiatry. 14, 1233981 (2023).
Choi, K. W., Kim, Y. K. & Jeon, H. J. Comorbid anxiety and depression: clinical and conceptual consideration and transdiagnostic treatment. Anxiety disorders: Rethinking Underst. recent. discoveries 219–235 (2020).
Chen, X. & Li, S. Serial mediation of the relationship between impulsivity and suicidal ideation by depression and hopelessness in depressed patients. BMC public. health. 23, 1457 (2023).
Karalunas, S. L., Antovich, D., Miller, N. & Nigg, J. T. Prospective prediction of developing internalizing disorders in ADHD. J. Child Psychol. Psychiatry. 64, 768–778 (2023).
Howard, M. C. & Hoffman, M. E. Variable-centered, person-centered, and person-specific approaches: Where theory meets the method. Organizational Res. Methods. 21, 846–876 (2018).
Schmiege, S. J., Masyn, K. E. & Bryan, A. D. Confirmatory latent class analysis: Illustrations of empirically driven and theoretically driven model constraints. Organizational Res. Methods. 21, 983–1001 (2018).
Lei, H. et al. Symptom patterns of the occurrence of depression and anxiety in a Japanese general adult population sample: a latent class analysis. Front. Psychiatry. 13, 808918 (2022).
Sun, J. et al. Symptoms of depression and anxiety in Chinese adolescents: heterogeneity and associations with executive function. BMC psychiatry. 23, 410 (2023).
Yüksel, A. & Bahadir-Yilmaz, E. Relationship between depression, anxiety, cognitive distortions, and psychological well‐being among nursing students. Perspect. Psychiatr. Care. 55, 690–696 (2019).
Cramer, A. O., Waldorp, L. J., Van Der Maas, H. L. & Borsboom, D. Comorbidity: A network perspective. Behav. Brain Sci. 33, 137–150 (2010).
Fried, E. I. et al. Mental disorders as networks of problems: a review of recent insights. Soc. Psychiatry Psychiatr. Epidemiol. 52, 1–10 (2017).
Cai, H. et al. A network model of depressive and anxiety symptoms: a statistical evaluation. Mol. Psychiatry. 29, 767–781 (2024).
Garabiles, M. R., Lao, C. K., Xiong, Y. & Hall, B. J. Exploring comorbidity between anxiety and depression among migrant Filipino domestic workers: a network approach. J. Affect. Disord. 250, 85–93 (2019).
Ruan, Q. N., Chen, Y. H. & Yan, W. J. A network analysis of difficulties in emotion regulation, anxiety, and depression for adolescents in clinical settings. Child Adolesc. Psychiatry Mental Health. 17, 29 (2023).
Peng, P. & Liao, Y. Six addiction components of problematic social media use in relation to depression, anxiety, and stress symptoms: a latent profile analysis and network analysis. BMC psychiatry. 23, 321 (2023).
Borsboom, D. A network theory of mental disorders. World psychiatry. 16, 5–13 (2017).
Demır, Ü. F. & Bozkurt, O. Effects of perceived social support, depression and anxiety levels on migraine. Archives Neuropsychiatry. 57, 210 (2020).
Zhang, X. et al. Does depression affect the association between prosocial behavior and anxiety? A cross-sectional study of students in China. Front. public. health. 11, 1274253 (2023).
Bertollo, A. G., Leite Galvan, A. C., Mingoti, D., Dallagnol, M. E. & Ignácio, Z. M. C. Impact of COVID-19 on Anxiety and Depression-Biopsychosocial Factors. CNS & Neurological Disorders-Drug Targets (Formerly Current Drug Targets-CNS & Neurological Disorders). 23, 122–133 (2024).
Kroenke, K., Spitzer, R. L. & Williams, J. B. The PHQ-9: validity of a brief depression severity measure. J. Gen. Intern. Med. 16, 606–613. 10.1046/j.1525-1497.2001.016009606.x (2001).
Spitzer, R. L., Kroenke, K., Williams, J. B. & Löwe, B. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch. Intern. Med. 166, 1092–1097 (2006).
Cohen, S., Kamarck, T. & Mermelstein, R. A global measure of perceived stress. J. Health Soc. Behav. 385–396 (1983).
Soto, C. J. & John, O. P. The next Big Five Inventory (BFI-2): Developing and assessing a hierarchical model with 15 facets to enhance bandwidth, fidelity, and predictive power. J. Personal. Soc. Psychol. 113, 117 (2017).
Richang, Z. Psychological diagnosis of college students (Shandong Education, 1999).
Schwarzer, R. Generalized self-efficacy scale. Measures in health psychology: A user’s portfolio. Causal and control beliefs/Nfer-Nelson (1995).
Wang, C., Hu, Z. F. & Liu, Y. Evidences for reliability and validity of the Chinese version of general self-efficacy scale. Chin. J. Appl. Psychol. 7, 37–40 (2001).
Blumenthal, J. A. et al. Social support, type A behavior, and coronary artery disease. Psychosom. Med. 49, 331–340 (1987).
Foti, R. J., Bray, B. C., Thompson, N. J. & Allgood, S. F. Know thy self, know thy leader: Contributions of a pattern-oriented approach to examining leader perceptions. Leadersh. Q. 23, 702–717 (2012).
Morin, A. J., Morizot, J., Boudrias, J. S. & Madore, I. A multifoci person-centered perspective on workplace affective commitment: A latent profile/factor mixture analysis. Organizational Res. Methods. 14, 58–90 (2011).
Nylund, K. L., Asparouhov, T. & Muthén, B. O. Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Struct. equation modeling: multidisciplinary J. 14, 535–569 (2007).
Tofighi, D. & Enders, C. K. Identifying the correct number of classes in growth mixture models. Advances in latent variable mixture models. 317 (2008). (2007).
Lubke, G. & Muthén, B. O. Performance of factor mixture models as a function of model size, covariate effects, and class-specific parameters. Struct. Equation Modeling: Multidisciplinary J. 14, 26–47 (2007).
Friedman, J., Hastie, T. & Tibshirani, R. Sparse inverse covariance estimation with the graphical lasso. Biostatistics. 9, 432–441 (2008).
Epskamp, S. & Fried, E. I. A tutorial on regularized partial correlation networks. Psychol. Methods. 23, 617 (2018).
Opsahl, T., Agneessens, F. & Skvoretz, J. Node centrality in weighted networks: Generalizing degree and shortest paths. Social networks. 32, 245–251 (2010).
Jones, P. J., Ma, R. & McNally, R. J. Bridge centrality: a network approach to understanding comorbidity. Multivar. Behav. Res. 56, 353–367 (2021).
Epskamp, S., Borsboom, D. & Fried, E. I. Estimating psychological networks and their accuracy: A tutorial paper. Behav. Res. Methods. 50, 195–212 (2018).
Podsakoff, P. M., MacKenzie, S. B. & Podsakoff, N. P. Sources of method bias in social science research and recommendations on how to control it. Ann. Rev. Psychol. 63, 539–569 (2012).
Curran, E., Rosato, M., Ferry, F. & Leavey, G. Prevalence and factors associated with anxiety and depression in older adults: Gender differences in psychosocial indicators. J. Affect. Disord. 267, 114–122 (2020).
Kong, Z. et al. Somatic symptoms mediate the association between subclinical anxiety and depressive symptoms and its neuroimaging mechanisms. BMC psychiatry. 22, 835 (2022).
Heeren, A., Hanseeuw, B., Cougnon, L. A. & Lits, G. Excessive worrying as a central feature of anxiety during the first COVID-19 lockdown-phase in Belgium: insights from a network approach. Physiol. Belgica. 61, 401 (2021).
Wang, Y., Hu, Z., Feng, Y., Wilson, A. & Chen, R. Changes in network centrality of psychopathology symptoms between the COVID-19 outbreak and after peak. Mol. Psychiatry. 25, 3140–3149 (2020).
Wang, B., Yuan, D., Zhong, X., Yang, F. & Fu, H. Family function, anxiety and depression in adults with disabilities: a network analysis. Front. public. health. 11, 1181203 (2023).
Bian, Z. et al. Associations between anxiety, depression, and personal mastery in community-dwelling older adults: a network-based analysis. BMC psychiatry. 24, 192 (2024).
Rapuano, M. et al. Spaces for relaxing, spaces for recharging: How parks affect people's emotions. J. Environ. Psychol. 81, 101809 (2022).
Can, Y. S. et al. in Healthcare. 100 (MDPI).
Chen, X. et al. The comorbidity of depression and anxiety symptoms in tinnitus sufferers: a network analysis. Brain Sci. 13, 583 (2023).
Teuber, M., Leyhr, D. & Sudeck, G. Physical activity improves stress load, recovery, and academic performance-related parameters among university students: a longitudinal study on daily level. BMC Public. Health. 24, 598 (2024).
Hassan, M. & Asaad, T. Tension-type headache, its relation to stress, and how to relieve it by cryotherapy among academic students. Middle East. Curr. Psychiatry. 27, 1–11 (2020).
McLeish, A. C., Walker, K. L. & Hart, J. L. Emotion Dysregulation and E-Cigarette Expectancies among College Student E-Cigarette Users. Subst. Use Misuse. 59, 920–927 (2024).
Mattos, L. A., Schmidt, A. T., Henderson, C. E. & Hogue, A. Therapeutic alliance and treatment outcome in the outpatient treatment of urban adolescents: The role of callous–unemotional traits. Psychotherapy. 54, 136 (2017).
Campbell, S. & Osborn, T. L. Adolescent psychopathology and psychological wellbeing: a network analysis approach. BMC psychiatry. 21, 333 (2021).
Stuckey, H. L. et al. Personal accounts of the negative and adaptive psychosocial experiences of people with diabetes in the second Diabetes Attitudes, Wishes and Needs (DAWN2) study. Diabetes care. 37, 2466–2474 (2014).
Wu, D., Shi, Z., Wu, C., Sun, W. & Jin, G. Sex differences in symptom network structure of depression, anxiety, and self-efficacy among people with diabetes: a network analysis. Front. Public. Health. 12, 1368752 (2024).
Konac, D., Young, K. S., Lau, J. & Barker, E. D. Comorbidity between depression and anxiety in adolescents: Bridge symptoms and relevance of risk and protective factors. J. Psychopathol. Behav. Assess. 43, 583–596 (2021).
Tao, Y. et al. Centrality and bridge symptoms of anxiety, depression, and sleep disturbance among college students during the COVID-19 pandemic—a network analysis. Curr. Psychol. 43, 13897–13908 (2024).
Bellingtier, J. A., Mund, M. & Wrzus, C. The role of extraversion and neuroticism for experiencing stress during the third wave of the COVID-19 pandemic. Curr. Psychol. 42, 12202–12212 (2023).
Ruan, Q. N., Chen, C., Jiang, D. G., Yan, W. J. & Lin, Z. A network analysis of social problem-solving and anxiety/depression in adolescents. Front. Psychiatry. 13, 921781 (2022).
Cumsille, P. E. & Epstein, N. Family cohesion, family adaptability, social support, and adolescent depressive symptoms in outpatient clinic families. J. Fam. Psychol. 8, 202 (1994).
Wight, R. G., Botticello, A. L. & Aneshensel, C. S. Socioeconomic context, social support, and adolescent mental health: A multilevel investigation. J. Youth Adolesc. 35, 109–120 (2006).
Cai, H. et al. Network analysis of depressive and anxiety symptoms in adolescents during the later stage of the COVID-19 pandemic. Translational Psychiatry. 12, 98 (2022).
Kaiser, T., Herzog, P., Voderholzer, U. & Brakemeier, E. L. Unraveling the comorbidity of depression and anxiety in a large inpatient sample: Network analysis to examine bridge symptoms. Depress. Anxiety. 38, 307–317 (2021).
McElroy, E., Fearon, P., Belsky, J., Fonagy, P. & Patalay, P. Networks of depression and anxiety symptoms across development. J. Am. Acad. Child. Adolesc. Psychiatry. 57, 964–973 (2018).

No competing interests reported.

Download PDF

Reviews received at journal
11 Nov, 2024
Reviewers agreed at journal
30 Oct, 2024
Reviewers invited by journal
30 Oct, 2024
Editor assigned by journal
30 Oct, 2024
Editor invited by journal
15 Oct, 2024
Submission checks completed at journal
14 Oct, 2024
First submitted to journal
26 Sep, 2024

You are reading this latest preprint version

Heterogeneity of depression and anxiety among college students and related risk and protective factors: a network model

Status:

Version 1

Abstract

Figures

Introduction

Materials and Methods

Participants

Measurements

Depression and Anxiety

Risk and Protective Factors

Statistical Analysis

Results and Analysis

Common Method Bias Test

Descriptive Statistics

Latent Profiles of Depression and Anxiety Symptoms

Model Optimization

Naming Latent Classes

Network Analysis

Symptoms of Depression and Anxiety

Risk and Protective Factors

Discussion

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1