Depression and mental health in late adolescents: the sequential mediating roles of quality of life and sleep regularity

Abstract

Background

Depression is one of the most prevalent mood disorders during adolescence and has profound implications for mental health and overall well-being. However, the mechanisms through which depression influences adolescent mental health remain insufficiently understood, particularly regarding modifiable psychosocial factors. Quality of life (QL) and sleep regularity (SR) have been identified as important correlates of both depression (DPS) and mental health (MH), yet their sequential roles have rarely been examined within a single explanatory framework.

Methods

A cross-sectional survey was conducted among 487 late adolescents using validated self-report instruments measuring depression, quality of life, sleep regularity, and mental health. Data were analyzed using partial least squares structural equation modeling (PLS-SEM) to assess the measurement model, structural relationships, and mediating effects. Bootstrapping with 5,000 resamples was employed to test indirect and sequential mediation effects.

Results

Depression was found to be negatively associated with quality of life (β = −0.344, p < 0.001). Quality of life positively predicted sleep regularity (β = 0.276, p < 0.001), which in turn was positively associated with mental health (β = 0.232, p < 0.01). Mediation analyses revealed that quality of life partially mediated the relationship between depression and mental health (β = −0.144, p < 0.001) and between depression and sleep regularity (β = −0.095, p < 0.001). Furthermore, a significant sequential mediation effect was observed, indicating that depression indirectly influenced mental health through the combined pathways of quality of life and sleep regularity (β = −0.022, p = 0.001). The model demonstrated acceptable explanatory power and overall fit.

Conclusions

The findings indicate that depressive symptoms are statistically associated with lower positive mental health among late adolescents, both directly and indirectly through quality of life and sleep regularity. These results suggest that quality of life and sleep regularity represent potentially modifiable psychosocial and behavioral correlates within this association. However, given the cross-sectional design, the observed pathways should be interpreted as patterns of association rather than evidence of temporal or causal effects. Future longitudinal and intervention-based studies are needed to clarify the directionality and practical implications of these relationships.

Clinical trial registration

Not applicable.

Introduction

Adolescence is a critical period of development. Profound changes have taken place in physical, psychological and social relations. During this period, individuals are particularly prone to emotional disorders [1, 2]. DPS is one of the most common MH diseases affecting late adolescents worldwide, which is related to the impairment of emotional function [3, 4], academic difficulties [5], social withdrawal [6, 7], and increased risk of long-term mental illness [8, 9]. Epidemiological evidence shows that depressive symptoms in adolescence tend to persist into adulthood [10], emphasizing the importance of identifying early mechanisms that link depression with broader MH outcomes [9]. In this study, we use the term “late adolescence” to refer to individuals aged 18–23. This is a stage of development, which usually overlaps with adulthood, and is characterized by increased autonomy, identity consolidation and unstable lifestyle.

Mental health in late adolescence is a multidimensional structure including emotional well-being, cognitive function, behavior regulation and social adaptation [11, 12]. Positive mental health does not simply reflect the inadequacy of psychopathology, but rather the ability of individuals to maintain resilience, life satisfaction, and effective social functioning in the face of development challenges [13, 14]. Previous studies have established a robust association between depression and impaired MH in late adolescents [15, 16].

In late adolescents, quality of life has become an important psychosocial factor closely related to depression and mental health [17, 18]. Quality of life is an individual ‘s subjective evaluation of their physical, psychological and social functions in daily life, reflecting the overall adaptation of late adolescents to their development background [19]. Empirical studies have shown that depressive symptoms are closely related to the decline in quality of life [20, 21], manifested as reduced life satisfaction, impaired social relations, and reduced participation in recreational activities. At the same time, higher quality of life is associated with better psychological well-being, emotional stability and adaptive function [22].

Recent studies have further emphasized the importance of sleep regularity and sleep related patterns in adolescent mental health [23, 24]. Emerging evidence suggests that irregular sleep wake rhythms are associated with the trajectory of depressive symptoms and broader psychological functions over time [25]. In addition, contemporary research emphasizes that the interaction between psychosocial function, life satisfaction and behavior rhythm is related to mental health vulnerability [26]. However, despite these advances, few studies have explicitly integrated quality of life and leep regularity into a single sequential interpretation framework. Therefore, the potential interdependence between subjective life assessment and sleep related behavioral regulation is still not clear.

To address these gaps, this study proposes a sequential mediation model guided by theory, in which the quality of life and sleep regularity jointly explain the association between depressive symptoms and positive mental health in late adolescents aged 18–23. By integrating psychosocial (quality of life) and behavioral (sleep regularity) factors into a single structural framework, this study goes beyond the static and single mediation model, and provides a more complete explanatory framework to test the relationship between these structures, that is, how depressive symptoms are statistically related to broader mental health functions.

Specifically, using the partial least squares structural equation model (PLS-SEM), we examined: (a) the direct association between depression and mental health; (b) The mediating role of quality of life in the relationship between depression and sleep regularity; (c) The sequential mediating role of quality of life and sleep regularity in linking depressive symptoms with positive mental health. By clarifying these interrelationships, the purpose of this study is to help improve the developmental mental health model and identify potential modifiable psychosocial and behavior related early prevention strategies.

Literature review and research hypotheses

Depression and quality of life

The later stage of adolescence is a sensitive development window period for the occurrence and consolidation of emotional problems. Recent large-scale evidence shows that depression or depressive symptoms affect a considerable proportion of children and late adolescents worldwide and show an increasing trend [3, 4, 27], which highlights the urgency of identifying modifiable pathways for protection and well-being. In addition to symptom burden, depression is also associated with impaired psychosocial function and reduced quality of life [28]. QL reflects the individual ‘s subjective evaluation of daily function, life satisfaction and well-being in the field of life. In late adolescents, studies have shown a correlation between DPS and reduced QL [18, 29]. From the perspective of impaired function, depressive symptoms can weaken motivation, social participation and adaptive coping, thereby reducing satisfaction with daily life and perceived quality of life.

H1

Depression has an effect on quality of life.

Quality of life and sleep regularity

Sleep is a core behavioral system, which is closely related to the daily function of late adolescents. Although many studies have focused on sleep time and sleep quality, more and more attention has been focused on sleep regularity, reflecting the stability and consistency of sleep-wake time and pattern between days [30, 31]. Sleep regularity is important in theory because it reflects the relationship between daily behavior and circadian rhythms and is considered to be a potentially modifiable behavioral factor associated with mental health. Relevant studies have shown that irregular sleep patterns may be prospectively associated with the deterioration of depressive symptoms, emphasizing the development significance of establishing a stable sleep-wake rhythm [25, 32].

QL may be used as an upstream predictor of sleep regularity. Late adolescents with higher QL are more likely to maintain structured daily behavior, better role functioning, and healthier lifestyle habits - factors that support a consistent sleep-wake schedule. Conversely, the decrease in QL may reflect the disruption of daily structure and psychosocial resources, increasing vulnerability to irregular sleep patterns.

H2

Quality of life positively influences sleep regularity.

Sleep regularity and mental health

Positive mental health emphasizes emotional well-being, adaptive function and psychological resources [33], which can be evaluated by short verification tools such as the Positive Mental Health Scale. In this framework, sleep patterns are highly correlated, because stable sleep-wake rhythms support the core processes of emotion regulation, cognitive control, and stress recovery, which are the basis of mental health functions [30, 34].

Empirical evidence shows that irregular sleep is not only associated with concurrent mental health difficulties, but also may predict changes in depressive symptoms over time [35, 36], suggesting that sleep patterns can be used as a behavioral pathway to connect psychosocial function and mental health outcomes. Therefore, late adolescents who maintain a more regular sleep pattern will have better overall mental health.

H3

Sleep regularity positively influences mental health.

Indirect and sequential mediating mechanisms

Although depression is widely believed to be detrimental to adolescent MH [37], there is growing evidence that this relationship is not purely direct ; on the contrary, it is carried out through intermediate processes of function and behavior. QL is a theoretically meaningful mediator variable, because depressive symptoms usually first impair perceived function and life satisfaction [38], and then can shape downstream behavior and health-related routines. The findings of related studies have linked depression to reduced QL, supporting the possibility of QL as an intermediate theoretically consistent association [28, 39].

Furthermore, QL and sleep regularity are likely to operate in a sequential process. That is, depressive symptoms are statistically associated with lower quality of life, which is in turn associated with less stable daily routines and irregular sleep–wake patterns. Sleep irregularity is statistically linked to greater difficulties in emotional regulation and psychological functioning, thereby undermining overall mental health. Longitudinal evidence that sleep irregularity predicts prospective increases in depressive symptoms [40] further supports the developmental salience of sleep regularity as a mechanism-related construct [36]. Based on this reasoning, this study proposes both single mediation and sequential mediation pathways (See Fig. 1):

Fig. 1.

Hypothesized model

H4

Depression has an indirect effect on mental health through QL.

H5

Depression has an indirect effect on sleep regularity through QL.

H6

Depression indirectly influences mental health through the sequential mediating effects of quality of life and sleep regularity.

Methods

Participants

This study was conducted from December 2025 to investigate the effects of college depression, QL, and sleep regularity on mental health. In this study, a random sampling method was used to distribute and collect questionnaires. The research team first worked with teachers in each class to clarify the purpose and importance of the questionnaire to ensure that teachers were able to properly convey this information. Therefore, teachers explain the background and significance of the questionnaire to students in the classroom to enhance their willingness to participate. Teachers randomly selected students to participate in the survey, using random number generator and other methods to ensure the unpredictability and representativeness of the sample. After selecting students, the teacher publishes a questionnaire link on the class group chat, emphasizing that participation is optional. Participants are told that they may choose to participate at any time and that their response will be strictly guaranteed. All participants are free to participate after knowing the purpose of the study, and take preventive measures to protect the privacy of personal information. During the data collection, the wrong answers (such as incorrect answers or repeated submissions) were excluded, and 487 valid questionnaires were obtained.

The inclusion criteria are: (a) full time undergraduate enrollment; (b) Between 18 and 23 years of age; (c) Voluntary participation after reading the informed consent form. Exclusion criteria include incomplete questionnaire answers, patterned answers (for example, the same item answers) or repeated submissions determined by IP filtering. Although the random number generation method is used to implement the random selection procedure in the class, the samples are from a limited number of universities in Guangdong Province, which may limit generalizability. Participation was voluntary, and students experiencing higher or lower psychological distress may have been differentially motivated to participate, introducing potential self-selection bias.

As shown in Table 1, within the sample, 228 participants were male (46.8%) and 259 were female (53.2%); Most were between the ages of 18 and 20 (72.9%); Third-year undergraduate students constituted 46.2% (225 individuals), Sophomore students 38.4% (187 individuals), while freshman students were relatively fewer; by major category, humanities accounted for 25.05% (122 individuals), science and engineering 53.39% (260 individuals), and arts and design 21.56% (105 individuals).

Table 1.

Demographic characteristics

Demographic Characteristics	Category	Quantity	Proportion
Gender	Male	228	46.8%
	Female	259	53.2%
Age	18–20	355	72.9%
	21–23	130	26.7%
Grade	23 and above	2	0.41%
	Freshman	74	15.2%
Major	Sophomore	187	38.4%
	Third-year undergraduate	225	46.2%
	Humanities	122	25.05%
	Science and Engineering Arts and Design	260 105	53.39% 21.56%

Measures

All tools used in this study were self-report scales previously published and validated. No new questionnaire was developed in this study. The scales have been widely used in previous studies and have proved satisfactory psychometric properties. No items were deleted or modified in this study. If necessary, fine tune the wording without changing the meaning of the project to ensure that the context is suitable for college students. Detailed project sources and representative sample projects are provided below, and complete references are cited accordingly.

Depression

Depression was measured using a previously published and validated scale developed by Randall et al. [41]. A typical example is ‘I feel frustrated and depressed.’ The scale uses a 5-point Likert scale, with 1 = strongly disagree and 5 = strongly agree. The scale has good reliability and validity. In this study, the α coefficient is 0.952.

Quality of life

Quality of life was measured using a previously published and validated scale developed by Burckhardt & Anderson [42]. A typical example is ‘You are satisfied with your recreational activities.’ The scale has good reliability and validity. In this study, the α coefficient is 0.926.

Sleep regularity

Sleep regularity was measured using a previously published and validated scale developed by Yan et al. [43]. A typical example is ‘My nightly sleep duration is approximately the same.’ The scale has good reliability and validity. In this study, the α coefficient is 0.878.

Mental health

Mental health was measured using a previously published and validated scale developed by Lukat et al. [44]. It is often used to assess the overall state of college students’ emotions, cognition, behavior, and social function (e.g., “I live a happy life; I am confident to express my opinions”). Cronbach’s alpha was 0.916. The scale showed good reliability.

Procedure

In this study, we used the depression, quality of life, sleep regularity and mental health scales to collect data through online questionnaires. All subjects agreed to sign the informed consent form before starting to fill out the online questionnaire. After the participants completed all the items, the test data was automatically generated. The procedure was approved by the Ethics Committee of Guangdong University of Science and Technology.

Statistical analysis

This study uses SPSS 29.0 software for descriptive statistics on the data. Additionally, we will use SmartPLS 4.1 software for the measurement model and structural evaluation during the data analysis process. This work applies PLS-SEM. This survey chose this software for analysis because it has been successful in evaluating validity and reliability and confirming or rejecting hypotheses [45]. In terms of analytical advantages, partial least squares can simultaneously estimate the path coefficients of the specified model and the loadings of individual items. Consequently, it enables researchers to circumvent biased and inconsistent parameter estimates [46] and is applied to more advanced and complex models [47].

The present study employed partial least squares structural equation modeling (PLS-SEM) rather than covariance-based SEM (CB-SEM) for several methodological reasons. First, the primary aim of this study was predictive and explanatory rather than strictly confirmatory. Specifically, we sought to maximize the explained variance of positive mental health and to examine the statistical pathways linking depressive symptoms, quality of life, and sleep regularity within a variance-based framework. PLS-SEM is particularly well suited for prediction-oriented research in which the focus is on explaining endogenous construct variance (R²) rather than reproducing the empirical covariance matrix [48]. Second, although the proposed model is structurally parsimonious, it incorporates a sequential mediation process integrating psychosocial and behavioral constructs. PLS-SEM offers flexibility in estimating such process-oriented models and provides stable parameter estimates using bootstrapping procedures. Third, the data were collected using self-report Likert-type scales, which may not fully meet multivariate normality assumptions required by covariance-based SEM. As a variance-based and distribution-free approach, PLS-SEM imposes fewer distributional restrictions and is therefore appropriate for the present dataset.

Taken together, these considerations support the use of PLS-SEM as a suitable analytic strategy for addressing the predictive and exploratory objectives of this study.

Covariates and robustness checks. Given the observational and cross-sectional design, we examined whether key demographic variables could confound the estimated relationships. Specifically, gender, age, grade, and major were tested as covariates by adding direct paths from each covariate to the endogenous constructs in the structural model (particularly mental health). The inclusion of these covariates did not yield statistically significant effects, and the magnitude and significance of the hypothesized structural paths remained materially unchanged. Therefore, to preserve model parsimony and facilitate interpretability, the final model presented in the manuscript excludes these covariates. The conclusions are robust to the inclusion of demographic controls.

Mediation effects were interpreted as statistical partitioning of associations within a cross-sectional framework, without implying temporal precedence.

Results

Common method bias test

All variables in this study were measured through self-report questionnaires from the same respondents. To reduce the potential common method bias, the study adopted several procedural control measures during the data collection process. First, at the beginning of the questionnaire, the respondents were clearly informed that the study was only for academic purposes, all answers would be kept strictly confidential, and there were no “correct” or “wrong” answers, so as to reduce the social expectation effect. At the same time, respondents were encouraged to answer based on their true feelings, so as to improve the authenticity and reliability of the data. In addition, the presentation order of the questionnaire questions is randomized to reduce the systematic bias that may be caused by the order of the answers [49]. In terms of statistical control, the common method bias was tested using the full collinearity test in SmartPLS software based on Kock’s suggestion [50]. The results indicated that the variance inflation factor (VIF) values for all latent constructs were below the threshold of 3.3 [51], indicating that there is no significant common method bias in this study.

Measurement model

According to Table 2, all constructs show acceptable reliability and convergent validity. Specifically, the Cronbach’s α value was between 0.826 and 0.952, and the CR value was between 0.878 and 0.953, both exceeding the recommended threshold of 0.70. AVE values were between 0.576 and 0.778, higher than the minimum cutoff of 0.50, indicating satisfactory convergent validity. In addition, according to the standard, item factor loadings should exceed 0.7 [52]. Consequently, the overall convergent validity remains satisfactory [53]. The HTMT values in Table 3 ranged from 0.278 to 0.605, which are all below the conservative threshold of 0.85. This demonstrates that each construct is empirically distinct and meets the requirements of discriminant validity [52]. As indicated in Table 4, the square roots of the AVEs (diagonal values) were all greater than the inter-construct correlations (off-diagonal values). These results further confirm the discriminant validity of the measurement model [52].

Table 2.

Reliability and validity

Constructs	Items	Outer Loadings	Cronbach’α	CR	AVE
Depression	DPS 1 DPS 2 DPS 3 DPS 4 DPS 5 DPS 6 DPS 7	0.884 0.913 0.887 0.899 0.871 0.904 0.812	0.952	0.953	0.778
Mental Health	MH 1 MH 2 MH 3 MH 4 MH 5 MH 6 MH 7	0.824 0.702 0.820 0.753 0.781 0.799 0.780	0.893	0.916	0.610
Sleep Regularity	SR 1 SR 2 SR 3 SR 4	0.804 0.767 0.825 0.813	0.826	0.878	0.644
Quality of Life	QL 1 QL 2 QL 3 QL 4 QL 5 QL 6 QL 7 QL 8 QL 9 QL 10 QL 11	0.787 0.761 0.753 0.748 0.768 0.721 0.732 0.743 0.750 0.829 0.753	0.926	0.927	0.576

Table 3.

Discriminant validity (HTMT Criterion)

	DPS	MH	QL	SR
DPS
MH	0.401
QL	0.366	0.605
SR	0.052	0.391	0.278

Table 4.

Discriminant validity (Fornell-Larcker Criterion)

	DPS	MH	QL	SR
DPS	0.882
MH	-0.372	0.781
QL	-0.344	0.554	0.759
SR	-0.013	0.338	0.248	0.802

Structural model

The VIF values were examined to assess potential collinearity issues among the constructs. As shown in Table 5, the VIF values for DPS, MH, QL, and SR ranged between 1.000 and 1.216. All values were well below the commonly accepted threshold of 3.3, indicating that multicollinearity was not a concern in the structural model [52, 54].

Table 5.

Collinearity test

	DPS	MH	QL	SR
DPS		1.142	1.000	1.134
MH
QL		1.216		1.134
SR		1.072

The structural model was evaluated using bootstrapping with 5,000 resamples. As shown in Table 6; Fig. 2, depressive symptoms were moderately negatively associated with quality of life (β = −0.344, 95% CI [− 0.450, − 0.239]). In turn, quality of life demonstrated a moderate positive association with sleep regularity (β = 0.276, 95% CI [0.170, 0.381]), and sleep regularity was positively associated with mental health (β = 0.232, 95% CI [0.142, 0.320]).

Table 6.

Path hypothesis testing

Hypothesis	Original sample (β)	2.50%	97.50%	p	Results
DPS →QL	-0.344	-0.450	-0.239	0.000	Support
QL →SR	0.276	0.170	0.381	0.000	Support
SR → MH	0.232	0.142	0.320	0.000	Support

Fig. 2.

The SEM illustrates the relationships among DPS, QL, SR, and MH. Note: ***P < 0.001; **P < 0.01; *P < 0.05

The magnitude and direction of these effects were consistent with the hypothesized model. Results remained substantively unchanged when demographic covariates were included, indicating robustness of the structural estimates.

The model explained 37.6% of the variance in mental health (R² = 0.376), indicating moderate explanatory power. In contrast, the explained variance for quality of life (R² = 0.119) and sleep regularity (R² = 0.067) was modest, suggesting that additional determinants of these constructs likely exist beyond the present model. Model fit indices (SRMR = 0.053; NFI = 0.844) indicated acceptable fit (see Table 7).

Table 7.

Explanatory power and predictive relevance

	R ²	Model fit
MH	0.376	SRMR: 0.053
QL	0.119	NFI: 0.844
SR	0.067

Mediation analysis

Bootstrapping results supported the hypothesized mediation pathways (Table 8). Quality of life partially mediated the association between depression and mental health (β = −0.144, 95% CI [− 0.198, − 0.097], t = 5.431), as well as between depression and sleep regularity (β = −0.095, 95% CI [− 0.144, − 0.055], t = 4.215).

Table 8.

Mediating analysis

Relationship	Indirect Effect	2.5%	97.5%	t	p	Direct Effect	Type of Mediation
DPS → QL →MH	-0.144	-0.198	-0.097	5.431	0.000	-0.225	Partial mediation
DPS → QL →SR	-0.095	-0.144	-0.055	4.215	0.000	0.082	Partial mediation
DPS → QL →SR → MH	-0.022	-0.038	-0.011	3.251	0.001	-0.225	Partial mediation

The sequential indirect effect from depression to mental health via quality of life and sleep regularity was small but statistically significant (β = −0.022, 95% CI [− 0.038, − 0.011], t = 3.251), indicating partial mediation. The direct effect of depression on mental health remained significant after inclusion of mediators, supporting a partial mediation structure.

Discussion

Overview of key findings

It is important to distinguish between empirically supported associations observed in the present data and broader theoretical interpretations. While the statistical pathways were significant, the developmental ordering proposed in the model remains provisional and requires longitudinal validation.

The empirical results of this study largely support the proposed hypothesis, and the findings provide evidence for a statistically significant way to link depressive symptoms with positive mental health through quality of life and sleep regularity. However, given the modest proportion of variance explained in quality of life and sleep regularity, these constructs should be understood as partial contributors within a broader psychosocial context rather than comprehensive explanatory mechanisms. First, DPS has a significant negative impact on QL (H1), indicating that higher levels of depressive symptoms are related to late adolescents’ lower evaluation of daily function, social function and subjective well-being. This finding confirms the general erosive effect of depression on the overall functional status of adolescence. Second, quality of life significantly positively predicted sleep regularity (H2), indicating that higher life satisfaction and better daily function helped to maintain a more stable sleep-wake pattern. In addition, sleep regularity has a significant positive impact on MH (H3), supporting the association of sleep-wake rhythm in emotional regulation and psychological function maintenance.

Taken together, these findings support the core structural relationships proposed in the model and provide the foundation for examining the mediating processes in greater depth.

Interpretation of the sequential mediation model

In terms of mediating effect, QL plays a significant partial mediating role in the relationship between DPS and mental health (H4), and also plays a partial mediating role in the relationship between DPS and SR (H5). More importantly, the results support a significant sequential mediation path (H6), that is, the sequential mediation findings suggest that depressive symptoms are statistically associated with lower positive mental health partly through reduced quality of life and less regular sleep patterns. However, given the cross-sectional design, this pathway should be interpreted as a theoretically informed pattern of associations rather than definitive evidence of temporal ordering. Even after the mediating variables were included, the direct effect of DPS on MH was still significant, indicating that there was a partial mediating structure, highlighting the multi-path nature of the impact of depression on adolescent mental health. Importantly, the mediation findings should be understood as statistical decompositions of associations rather than confirmation of causal or temporal processes.

Integration with prior literature

In addition to confirming the previously established association, the findings of this study expand the existing literature by promoting a more complete understanding of how depressive symptoms are statistically associated with variations in adolescent mental health. Although previous studies have mainly documented a direct link between depressive symptoms and poor mental health outcomes [55–57], many of these studies rely on static models to capture relationships at a single analytical level. Although these methods provide information, they provide limited insight into the underlying psychosocial processes by which depression exerts its effects over time.

In particular, past studies have tended to study quality of life and sleep alone, either as a final result of depression, or as an independent mental health-related factor [58–60]. In this study, the quality of life and sleep regularity are put into the same model, which addresses prior conceptual fragmentation and provides a more integrated explanatory perspective.

In addition, his study also responds to recent calls in the literature to integrate psychosocial and behavioral determinants within a unified framework. When discussing the mental health of late adolescents, it is best to study the social psychological and behavioral factors together. Unlike the past ‘parallel correlation’ that simply juxtaposes sleep disorders and life dissatisfaction as depression, our proposed model emphasizes more on the dependence and cumulative effects between the two. This comprehensive view provides a more detailed explanation of how depressive symptoms occur in multiple functional areas and lead to sustained damage to mental health. Therefore, the results of this study not only verify the known correlation, but also provide empirical support for the dynamic multi-stage mechanism between depression and late adolescents mental health outcomes.

Theoretical implications

This study provides an important theoretical contribution to the literature on late adolescents depression and MH. First, this study defines mental health as a positive, multidimensional structure, not just a lack of psychopathology, which extends the traditional defect oriented adolescent depression model. This view is consistent with the contemporary mental health framework that emphasizes psychological function, well-being and adaptability, thus enriching the theoretical understanding of the mental health outcomes of late adolescents.

Secondly, this study has promoted the development of depression research by clarifying the mechanism of depression symptoms affecting mental health, rather than focusing on the direct connection. Quality of life and sleep regularity were identified as sequential mediating variables, which provided empirical support for the process oriented mechanism. It showed that depression was statistically correlated with lower mental health, partly through related psychosocial and behavioral pathways. This sequence structure offers a theoretically coherent framework for understanding how these constructs may be interrelated, which usually regards psychosocial factors and sleep-related variables as parallel or independent related factors [58, 61], but emphasizes their orderly and cumulative role in the development of adolescent mental health.

Third, the results emphasize the importance of adolescents’ subjective life evaluation, and contribute to the theory of developmental mental health. The quality of life is a proximal mechanism that links depressive symptoms with downstream behavioral regulation, suggesting that the perception of daily function and life satisfaction of late adolescents may represent an important psychosocial correlate associated with subsequent health-related behaviors. The results of this study show that the quality of life is not a basic driving factor, but a meaningful component, which plays a role in a broader emotional, environmental and behavioral impact [62]. This view supports the theoretical framework that late adolescents is a period in which subjective experience and self-evaluation have a strong impact on psychological and behavioral adjustment.

Fourth, the study included sleep regularity as behavioral mediators, which expanded the application of sleep related theories in mental health research. Unlike the traditional methods that mainly focus on sleep time or sleep quality [61, 63, 64], this study emphasizes the theoretical relevance of sleep regularity as a unique structure between psychosocial function and mental health outcomes. By positioning sleep regularity as behavior related associations in the model, the findings highlight their potential role in linking psychosocial functions and mental health. However, considering that sleep regularity is completely assessed through self-report, these conclusions should be carefully interpreted. It is necessary to combine objective or diary based measures with future research.

Practical implications

The results of this study have important practical significance for the prevention and management of late adolescent mental health problems, especially those related to depression. First, the identification of quality of life as a mediator indicates that interventions aimed at improving daily functioning and subjective well-being of late adolescents may be a promising area for supportive intervention in the development of depressive symptoms in MH. School and community-based programs that promote positive social relationships, engage in meaningful activities, and balance daily life may help buffer the adverse effects of depression at an early stage.

Second, the research results emphasize that sleep regularity is an operable and modifiable goal to promote mental health [30, 65]. Different from the clinical symptoms that usually require special treatment, sleep wake regularity can be solved by low-cost non pharmacological strategies, such as sleep hygiene education, consistent bedtime routine and structured schedule. In view of the important role of sleep regularity in linking psychosocial functions with mental health outcomes, promoting stable sleep regularity can be regarded as a potential auxiliary focus of existing adolescent MH interventions.

Third, the sequential mediation approach identified in this study suggests that comprehensive approaches may be beneficial, although further longitudinal and experimental evidence is required. Different from solving depression symptoms, life dissatisfaction or sleep problems in isolation [59, 66], practitioners and educators can benefit from comprehensive strategies of subjective well-being and behavior regulation. For example, programs aimed at improving the quality of life may indirectly improve sleep regularity, thereby bringing broader benefits to the mental health of late adolescents.

Limitations and future research directions

Although this study has made theoretical and empirical contributions, some limitations should be recognized and explicitly considered when interpreting the results. First, the cross-section design excludes any inference regarding temporal ordering or causal directionality. Although the sequential mediation model is theoretically grounded, the observed pathways represent contemporaneous statistical associations rather than developmental processes unfolding over time. As emphasized in the conclusion, these findings should be interpreted as associative patterns. Future studies need to adopt longitudinal, cross lag or intervention based designs to determine whether depressive symptoms prospectively affect quality of life, sleep regularity and mental health, or whether they involve a bidirectional processes.

Second, sleep regularity was assessed exclusively through self-report measures rather than objective indicators such as actigraphy or sleep diaries. As subjective sleep assessments may be influenced by mood state and recall bias, the observed associations may partially reflect shared method variance. Future studies should incorporate multi-method sleep assessment to validate and refine the proposed pathway.

Third, an additional conceptual consideration concerns the potential overlap among depression, quality of life, and positive mental health. Although discriminant validity was supported statistically through HTMT and Fornell–Larcker criteria, these constructs are theoretically proximal and partially overlapping in content. For example, depressive symptoms include emotional and cognitive components that may be inversely proportional to life satisfaction and well-being, while positive mental health reflects adaptive function and emotional balance, which conceptually intersects with all aspects of quality of life. Therefore, some observed associations may reflect shared conceptual domains rather than completely different psychological processes. In addition, since all structures are measured through the self-report of the same respondent at a single time point, shared method variance may have contributed to the magnitude of the relationships. Future research combining multiple information designs, clinician assessments or behavioral indicators will help to further disentangle conceptual proximity from the substantive causal process.

Fourth, the explained variance for quality of life (R² = 0.119) and sleep regularity (R² = 0.067) was modest. This suggests that important contextual and personal determinants such as socio-economic status, family environment, academic pressure, peer relationships, personality traits and lifestyle variability are not captured in the current model. Therefore, the quality of life and sleep regularity should be interpreted as partial mediating mechanisms in the broader psychosocial system, rather than a comprehensive explanation factor. Extending the model to include additional context predictions will enhance the integrity of interpretation.

Fifth, given the observational and cross-sectional nature of the data, the possibility of residual confounding cannot be excluded. Although demographic covariates were tested and did not alter the results, residual confounding due to unmeasured factors (e.g., socioeconomic status, family environment, academic stress) cannot be fully ruled out.

Finally, the sample was drawn from a limited number of universities within one province in China, which may constrain generalizability. Cultural norms, educational systems, and lifestyle patterns vary across regions and countries, and replication in diverse sociocultural contexts is warranted to assess the external validity of the model. The findings are specific to late adolescents enrolled in higher education, and may not generalize to early or mid-adolescent populations whose developmental contexts and sleep patterns differ substantially.

Conclusion

The findings indicate that depressive symptoms are statistically associated with lower positive mental health among late adolescents, both directly and indirectly through quality of life and sleep regularity. These results suggest that quality of life and sleep regularity represent potentially modifiable psychosocial and behavioral correlates within this association. However, given the cross-sectional design, the observed pathways should be interpreted as patterns of association rather than evidence of temporal or causal effects. Future longitudinal and intervention-based studies are needed to clarify the directionality and practical implications of these relationships.

Author contributions

Conceptualization: Y.Y., Q.Y., T.L. Y.F., J.D.; methodology: Y.Y., Q.Y., T.L. Y.F., J.D.; formal analysis: Y.Y., Q.Y., T.L. Y.F., J.D.; investigation: Y.Y., Q.Y., T.L. Y.F., J.D.; resources: Y.Y., Q.Y., T.L. Y.F., J.D.; writing—preparation of the original draft: Y.Y., Q.Y., T.L. Y.F., J.D.; writing—review & editing: Y.Y., Q.Y., T.L. Y.F., J.D. Each author has read and approved the published version of the manuscript. The co-first authors of this publication are Yanqing Yan and Qiuxian Ye.

Funding

1. The author(s) declare that financial support was received for the research and/or publication of this article. This research was funded by the Doctoral Research Start-up Program of Guangdong University of Science and Technology. The project, titled “Construction Mode and Practice of Applied Talents Training in University Sports” (grant no. GKY-2024BSQDW-84). 2. Guangdong Province Youth Campus Football and School Sports High Quality Development Special Topic: Research on Video Analysis System Based on AI Model Training to Improve Youth Football Technical and Tactical Ability (Project No.: 25SXZPT41).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in strict compliance with the standards of the 1964 Declaration of Helsinki. As the research involved human participants, it was reviewed and approved by the Academic Ethics and Morality Committee of Guangdong University of Science and Technology (Approval number: 2025020756). Before distributing the questionnaires, researchers informed participants about the survey’s purpose, background, duration, and intended use of the questionnaires. Questionnaires were distributed only after obtaining participants’ consent. The first page of the questionnaire included an informed consent form. Participants who objected to any part of the consent form could select the “Disagree” option to terminate their participation in the survey.

Consent for publication

Not applicable.

Informed Consent

Informed consent was obtained from all participants prior to their participation in the study. In line with national regulations and institutional policies, written (signed) consent was not required. Instead, an online consent process was embedded at the beginning of the questionnaire. Participants were presented with a detailed instruction page outlining: (i) voluntary participation and the right to withdraw at any time without consequence; (ii) confidentiality and anonymity, with no identifiable information collected; (iii) the academic-only purpose of data use; (iv) data protection measures, including passwordsecured local storage accessible only to the research team; and (v) that the consent interface had been pilot-tested for clarity and accessibility. Only those who clicked “Agree and Continue” were able to proceed.

Competing interests

The authors declare no competing interests.