Reciprocal relationship among perceived sleep quality, smartphone overdependence, and depression in Korean adolescents: a cross-lagged analysis using national big data Authors: Won-Oak Oh, Yoo-Jin Heo Published: 2025 ABSTRACT Adolescent depression is a critical public health concern, with growing links to excessive smartphone use and perceived poor sleep quality. However, the reciprocal relationships among these factors remain poorly understood. This longitudinal study examined the directional and temporal pathways linking smartphone overdependence, sleep quality, and depressive symptoms over time. Data were drawn from the Korea Children and Youth Panel Survey, including 2,052 adolescents who completed three consecutive waves from 2020 to 2022. Smartphone overdependence was measured using the Smartphone Addiction Self-Diagnosis Scale, perceived sleep quality was assessed with a single item developed by the National Institute of South Korea, and depressive symptoms were evaluated using the 13-item Brief Psychiatric Diagnostic Test. An autoregressive cross-lagged panel model was used to evaluate temporal and predictive relationships among these variables. The analysis revealed that poor sleep quality consistently predicted later smartphone overdependence and depressive symptoms. Depressive symptoms also predicted declines in sleep quality, indicating a bidirectional relationship. These findings highlight the importance of sleep as a central target for integrated interventions and may inform broader mental health policies aimed at improving adolescent well-being.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-33361-4. Introduction Depression is a profound and persistent form of psychological distress that can severely impair an individual’s physical and social functioning, affecting not only the person but also their family and broader community1,2. Owing to these characteristics, depression is considered one of the most serious global mental health concerns1. In late adolescence, approximately 25% of individuals experience at least one depressive episode3. In South Korea, 27.7% of adolescents reported experiencing feelings of depression in the past 12 months4. Globally, approximately 4.7% of adolescents are diagnosed with depressive disorders, and their prevalence continues to increase5,6. Adolescent depression directly influences academic achievement and school attendance7,8 and is associated with serious outcomes such as social withdrawal, increased suicide risk, and problematic substance use9,10. Adolescence is a critical developmental period characterized by structural and functional changes in the prefrontal cortex and limbic system11. Specifically, the limbic system, responsible for emotional processing and reward sensitivity, matures early, while the prefrontal cortex, responsible for executive and regulatory functions, continues to develop throughout adolescence11. This maturational asynchrony contributes to emotional dysregulation, impulse-control difficulties, and increased vulnerability to mental health problems12,13. Many cases of depression start during childhood and escalate sharply during adolescence14,15, with a high risk of becoming chronic during adulthood15. Despite sustained attention to the issue16, the key factors for preventing and treating adolescent depression remain insufficiently understood, and the problem has yet to be effectively resolved2,17. Consequently, here is an urgent need to comprehensively examine the contributing factors and mechanisms underlying adolescent depression, including its predictive relationships with key determinants of adolescent health. Among these contributing factors, increasing attention has been directed toward smartphone overdependence. Smartphones are ubiquitous among adolescents and positively affect their psychological and social development by facilitating access to information and promoting social interaction18,19. However, excessive smartphone use has been linked to a range of cognitive, emotional, and physical problems, including impaired cognitive functioning, difficulties with emotional regulation, impulsivity, low self-esteem, insomnia, and anxiety20,21. Consequently, smartphone overdependence has emerged as a key factor in adolescent depression, necessitating a comprehensive and longitudinal investigation. Perceived sleep quality during adolescence has also been identified as a critical factor influencing mental health. Poor sleep quality is reportedly associated with an increased risk of depression, anxiety, and difficulties with impulse control22,23. Adolescents are often chronically sleep deprived due to academic stress, heavy study loads, and a biologically driven late chronotype typical of this developmental period24–26. Despite the longstanding recognition of adolescent sleep problems, effective solutions remain elusive27. Among various influencing factors, increasing evidence suggests that both smartphone overdependence and depression are closely linked to sleep quality28–30. Understanding the temporal and directional relationships among these variables may provide critical insights for developing targeted interventions that effectively promote adolescent mental health. Recent research has increasingly explored associations between sleep quality, depressive symptoms, and smartphone overdependence, suggesting that these domains are key emotional and behavioral factors underlying adolescents’ psychological stability and healthy development. Longitudinal studies have demonstrated a relationship between general smartphone use, sleep, and depression in adolescents31. Extensive research has established a link between higher overall smartphone use and compromised sleep quality as well as elevated depressive symptoms in adolescents. However, evidence specifically addressing smartphone overdependence in this population remains relatively limited32–36. Moreover, the majority of existing studies have employed a cross-sectional design35, which renders it difficult to identify temporal or directional relationships among these variables. Several recent longitudinal studies have addressed these gaps. For example, Zhou et al.37 investigated bidirectional relationships between depressive symptoms and smartphone overdependence over a six-month interval in Chinese adolescents. Crowhurst and Hosseinzadeh38 systematically reviewed longitudinal evidence linking smartphone overdependence with mental health outcomes. Similarly, Touchette et al.39 tracked sleep trajectories from childhood to adolescence, demonstrating that early sleep patterns predict later depressive symptoms. Vazsonyi et al.40 conducted four repeated assessments over 1.5 years and reported negative longitudinal associations between sleep quality and progression of depressive and anxiety symptoms. While these studies provide valuable longitudinal evidence, the number of studies specifically examining the relationship between sleep quality and smartphone overdependence remains limited. Additionally, no prior research has simultaneously examined the longitudinal interplay among sleep quality, smartphone overdependence, and depressive symptoms. Depression, sleep quality, and smartphone overdependence are closely interrelated and influence adolescent mental health through complex underlying mechanisms35. Therefore, simultaneously assessing the relationships among these three variables is essential for a comprehensive understanding and for developing effective prevention and intervention strategies. This study addresses this gap by investigating concurrent and dynamic relationships over time, offering new insights into adolescent health. Adolescence is a critical developmental stage during which behavioral patterns and emotional responses are formed13, and health-related behaviors established during this period have lasting implications for adulthood41. Accordingly, this study examined reciprocal relationships between perceived sleep quality, depressive symptoms, and smartphone overdependence across three waves of data from middle-school adolescents. Using an autoregressive cross-lagged (ARCL) panel analysis, we simultaneously estimated autoregressive effects and cross-lagged paths to assess temporal changes and directional relationships between the study variables. This longitudinal approach addresses limitations of predominantly cross-sectional prior research and offers empirical evidence to inform interventions targeting adolescents’ emotional and behavioral health. Methods Aims This three-wave longitudinal study aimed to clarify the temporal stability and reciprocal associations among smartphone overdependence, perceived sleep quality, and depression in Korean adolescents. Grounded in prior empirical findings, this study addressed the following research questions: (1) What is the reciprocal relationship between smartphone overdependence and perceived sleep quality? (2) What is the reciprocal relationship between smartphone overdependence and depression? (3) What is the reciprocal relationship between perceived sleep quality and depression? Study design and population This study employed a national longitudinal descriptive design, using secondary data from the Korea Children and Youth Panel Survey 2018 (KCYPS 2018) conducted by the National Youth Policy Institute. The KCYPS is an ongoing longitudinal study initiated in 2018; it comprehensively assesses various aspects of child and adolescent growth and development in South Korea. To ensure representativeness, the survey adopts a multistage stratified cluster sampling method. For this study, data were drawn from a middle school cohort that followed a sample of first-year middle school students (Grade 7). Specifically, data from the third (2020, corresponding to Time 1), fourth (2021, corresponding to Time 2), and fifth (2022, corresponding to Time 3) waves were used, corresponding to Grades 9 (n = 2,384), 10 (n = 2,265), and 11 (n = 2,252), respectively. The final analytical sample included 2,052 adolescents who participated in all three waves and provided complete responses for the following key variables: smartphone overdependence, sleep quality, and depression. Measurement variables Smartphone overdependence Smartphone overdependence was assessed using the Smartphone Addiction Self-Diagnosis Scale developed by Kim et al.42. This 15-item scale consists of four categories: daily life disturbance, preference for the virtual world, withdrawal, and tolerance. Each item is rated on a four-point Likert scale ranging from 1 (not at all true) to 4 (very true), with higher scores indicating greater dependence. Cronbach’s alpha of this scale for this study was 0.877. Perceived sleep quality perceived Sleep quality was assessed using a single item developed by the National Youth Policy Institute in 2018, asking respondents whether they usually slept well on weekdays. Responses were rated on a four-point Likert scale ranging from 1 (very poor sleep) to 4 (very good sleep), with higher scores indicating better perceived sleep quality. Depression Depression was assessed using a 10-item scale adapted from the 13-item Brief Psychiatric Diagnostic Test developed by Kim et al.43, with three items excluded for developmental appropriateness. The scale includes three factors: emotional distress (e.g., feeling overwhelmed, frequent crying, and suicidal thoughts); excessive worry and lack of hope for the future; and experiences of loneliness, low energy, and persistent sadness. Each item is rated on a four-point Likert scale from 1 (not at all true) to 4 (very true), with higher scores indicating more severe depressive symptoms. Cronbach’s alpha of the scale for this study was 0.891. Data analysis Data analyses were performed using SPSS 24.0 and AMOS 26.0 (IBM Corp., Armonk, NY). Descriptive statistics were computed to summarize the adolescents’ sociodemographic characteristics and key study variables. Pearson’s correlation coefficients were calculated to examine associations among the study variables at each time point. To assess the directional and reciprocal effects of smartphone overdependence, perceived sleep quality, and depression over time, we conducted ARCL modeling using AMOS. The ARCL approach is well-suited for longitudinal data analysis, as it allows for estimating both the stability of each variable over time (autoregressive effects) and the cross-lagged influences among variables44. Before testing cross-lagged associations, a series of model comparisons evaluated key ARCL assumptions, including invariance of measurement loadings, structural paths, and error covariances over time. Measurement loadings confirm that indicators have consistent meaning across waves, enabling valid cross-time comparisons. Structural paths assess whether autoregressive and cross-lagged relationships remain stable. Error covariance assesses whether correlations among measurement errors for the same variable are consistent, indicating that unsystematic influences operate similarly across time45. Invariance was considered established if adding constraints did not significantly worsen model fit during the stepwise testing process, ensuring the stability of the model’s structure and relationships. A total of 13 nested models were tested sequentially, beginning with a baseline model with freely estimated parameters, followed by stepwise constraints for measurement, structural path, and error covariance invariance. Model comparisons were conducted based on changes in chi-square values (Δχ²) and degrees of freedom, while also considering changes in the comparative fit index (ΔCFI) due to the chi-square test’s sensitivity to sample size. Following Cheung and Rensvold46, a ΔCFI of 0.01 or less was interpreted as evidence of invariance. The analysis began with a configural model with no constraints (Model 1). Factor loading invariance was then tested for smartphone overdependence (Model 2), perceived sleep quality (Model 3), and depression (Model 4). Autoregressive path invariance was also tested for smartphone overdependence (Model 5), perceived sleep quality (Model 6), and depression (Model 7). Cross-lagged path constraints were applied sequentially to test the associations between smartphone overdependence and perceived sleep quality (Model 8), smartphone overdependence and depression (Model 9), perceived sleep quality and smartphone overdependence (Model 10), perceived sleep quality and depression (Model 11), depression and smartphone overdependence (Model 12), and depression and perceived sleep quality (Model 13). Finally, residual covariances between latent variables were constrained to be equal over time. Model fit was evaluated using several commonly accepted indices: the chi-square (χ²) statistic, CFI, Tucker–Lewis index (TLI), and root mean square error of approximation (RMSEA). A good model fit was indicated by CFI and TLI values above 0.90, and an acceptable model fit was indicated by values exceeding 0.8047,48. RMSEA values below 0.08 were regarded as indicating an acceptable model fit49. A significance level of 0.05 was used for all statistical tests. Ethical consideration This study was approved as exempt from review by the Institutional Review Board of Korea University (IRB No. KUIRB-2025-0317-01) due to the use of publicly available secondary data. The data were fully anonymized and de-identified prior to access by the authors, and no personal identifying information was used. Therefore, informed consent from individual participants was not required. This study was conducted in accordance with the ethical standards of the Declaration of Helsinki. Results Preliminary analyses A total of 2,052 adolescents who completed all three waves and responded to all key variables—smartphone overdependence, perceived sleep quality, and depression—were included in the final analysis. The descriptive statistics for each variable are presented in Table 1. Skewness (–0.34 to 0.72) and kurtosis (–0.48 to 0.55) values were within acceptable ranges (± 2 for skewness and ± 7 for kurtosis), indicating that the normality assumption was met50. Table 1Descriptive statistics for the variables (N = 2052).Variables123456789SO T11SO T20.48**1SO T30.46**0.49**1SQ T1− 0.14**− 0.12**− 0.11**1SQ T2− 0.06**− 0.13**− 0.12**0.35**1SQ T3− 0.06**− 0.06**− 0.06**0.28**0.28**1DP T10.33**0.22**0.17**− 0.26**− 0.13**− 0.13**1DP T20.20**0.37**0.22**− 0.16**− 0.21**− 0.11*0.48**1DP T30.10**0.16**0.26**− 0.15**− 0.13**− 0.20**0.37**0.44**1M(SD) = Mean (Standard Deviation). Table 1Descriptive statistics for the variables (N = 2052).Variables123456789SO T11SO T20.48**1SO T30.46**0.49**1SQ T1− 0.14**− 0.12**− 0.11**1SQ T2− 0.06**− 0.13**− 0.12**0.35**1SQ T3− 0.06**− 0.06**− 0.06**0.28**0.28**1DP T10.33**0.22**0.17**− 0.26**− 0.13**− 0.13**1DP T20.20**0.37**0.22**− 0.16**− 0.21**− 0.11*0.48**1DP T30.10**0.16**0.26**− 0.15**− 0.13**− 0.20**0.37**0.44**1M(SD) = Mean (Standard Deviation). Descriptive statistics for the variables (N = 2052). M(SD) = Mean (Standard Deviation). As presented in Table 2, all variables were significantly correlated, both within and across time points. Smartphone overdependence was consistently negatively associated with perceived sleep quality and positively associated with depressive symptoms, while perceived sleep quality was consistently negatively associated with depressive symptoms. Table 2Pearson correlations among smartphone overdependence, sleep quality, and depression at T1, T2, and T3.EffectBSEβSO(T1) → SO(T2)0.4950.0180.523***DP(T1) → DP(T2)0.4590.0160.503***SQ(T1) → SQ(T2)0.2950.0150.293***SO(T1) → SQ(T2)0.0170.0310.01SQ(T1) → DP(T2)− 0.0290.012− 0.036*SD(T1) → DP(T2)0.0030.0240.002SQ(T1) → SO(T2)− 0.0230.009− 0.041**DP(T1) → SQ(T2)− 0.0670.02− 0.06***DP(T1) → SO(T2)0.0110.0110.018SO(T2) → SO(T3)0.4950.0180.501***DP(T2) → DP(T3)0.4590.0160.456***SQ(T2) → DP(T3)− 0.0290.012− 0.035*SO(T2) → DP(T3)0.0030.0240.002SQ(T2) → SO(T3)− 0.0230.009− 0.042**DP(T2) → SO(T3)0.0110.0110.017SQ(T2) → SQ(T3)0.2950.0150.301***SO(T2) → SQ(T3)0.0170.0310.009DP(T2) → SQ(T3)− 0.0670.02− 0.056***SO = Smartphone Overdependence; SQ = Sleep Quality; DP = Depression; T1: Time 1, T2: Time 2, T3: Time 3; **p < .01.. Table 2Pearson correlations among smartphone overdependence, sleep quality, and depression at T1, T2, and T3.EffectBSEβSO(T1) → SO(T2)0.4950.0180.523***DP(T1) → DP(T2)0.4590.0160.503***SQ(T1) → SQ(T2)0.2950.0150.293***SO(T1) → SQ(T2)0.0170.0310.01SQ(T1) → DP(T2)− 0.0290.012− 0.036*SD(T1) → DP(T2)0.0030.0240.002SQ(T1) → SO(T2)− 0.0230.009− 0.041**DP(T1) → SQ(T2)− 0.0670.02− 0.06***DP(T1) → SO(T2)0.0110.0110.018SO(T2) → SO(T3)0.4950.0180.501***DP(T2) → DP(T3)0.4590.0160.456***SQ(T2) → DP(T3)− 0.0290.012− 0.035*SO(T2) → DP(T3)0.0030.0240.002SQ(T2) → SO(T3)− 0.0230.009− 0.042**DP(T2) → SO(T3)0.0110.0110.017SQ(T2) → SQ(T3)0.2950.0150.301***SO(T2) → SQ(T3)0.0170.0310.009DP(T2) → SQ(T3)− 0.0670.02− 0.056***SO = Smartphone Overdependence; SQ = Sleep Quality; DP = Depression; T1: Time 1, T2: Time 2, T3: Time 3; **p < .01.. Pearson correlations among smartphone overdependence, sleep quality, and depression at T1, T2, and T3. SO = Smartphone Overdependence; SQ = Sleep Quality; DP = Depression; T1: Time 1, T2: Time 2, T3: Time 3; **p < .01.. Cross-lagged analysis of smartphone overdependence, perceived sleep quality, and depression The model fit indices for all 13 nested models demonstrated reasonable fit, with CFI values exceeding 0.90, TLI values above 0.88, and RMSEA values ranging from 0.065 to 0.067 (Supplementary Table S1). Sequential model comparisons between Models 1 and 13 revealed no ΔCFI values exceeding 0.01, indicating that measurement invariance, path invariance, and error covariance invariance were all satisfied across time points. Accordingly, Model 13 was selected as the final model. The results for the final model are presented in Table 3 and illustrated in Fig. 1. All three variables exhibited significant autoregressive stability over time. Specifically, smartphone overdependence at an earlier time point significantly predicted smartphone overdependence at the subsequent time point (β = 0.523, p < .001; β = 0.501, p < .001). Similarly, perceived sleep quality (β = 0.293, p < .001; β = 0.301, p < .001) and depressive symptoms (β = 0.503, p < .001; β = 0.456, p < .001) significantly predicted themselves over time, indicating temporal stability for all constructs. Table 3Regression weights for cross-lagged SEM model.EffectBSEβSO(T1) → SO(T2)0.4950.0180.523***DP(T1) → DP(T2)0.4590.0160.503***SQ(T1) → SQ(T2)0.2950.0150.293***SO(T1) → SQ(T2)0.0170.0310.01SQ(T1) → DP(T2)− 0.0290.012− 0.036*SD(T1) → DP(T2)0.0030.0240.002SQ(T1) → SO(T2)− 0.0230.009− 0.041**DP(T1) → SQ(T2)− 0.0670.02− 0.06***DP(T1) → SO(T2)0.0110.0110.018SO(T2) → SO(T3)0.4950.0180.501***DP(T2) → DP(T3)0.4590.0160.456***SQ(T2) → DP(T3)− 0.0290.012− 0.035*SO(T2) → DP(T3)0.0030.0240.002SQ(T2) → SO(T3)− 0.0230.009− 0.042**DP(T2) → SO(T3)0.0110.0110.017SQ(T2) → SQ(T3)0.2950.0150.301***SO(T2) → SQ(T3)0.0170.0310.009DP(T2) → SQ(T3)− 0.0670.02− 0.056***SO = Smartphone Overdependence; SQ = Sleep Quality; DP = Depression; *p < 0.05 **p < 0.01, ***p < 0.001. Table 3Regression weights for cross-lagged SEM model.EffectBSEβSO(T1) → SO(T2)0.4950.0180.523***DP(T1) → DP(T2)0.4590.0160.503***SQ(T1) → SQ(T2)0.2950.0150.293***SO(T1) → SQ(T2)0.0170.0310.01SQ(T1) → DP(T2)− 0.0290.012− 0.036*SD(T1) → DP(T2)0.0030.0240.002SQ(T1) → SO(T2)− 0.0230.009− 0.041**DP(T1) → SQ(T2)− 0.0670.02− 0.06***DP(T1) → SO(T2)0.0110.0110.018SO(T2) → SO(T3)0.4950.0180.501***DP(T2) → DP(T3)0.4590.0160.456***SQ(T2) → DP(T3)− 0.0290.012− 0.035*SO(T2) → DP(T3)0.0030.0240.002SQ(T2) → SO(T3)− 0.0230.009− 0.042**DP(T2) → SO(T3)0.0110.0110.017SQ(T2) → SQ(T3)0.2950.0150.301***SO(T2) → SQ(T3)0.0170.0310.009DP(T2) → SQ(T3)− 0.0670.02− 0.056***SO = Smartphone Overdependence; SQ = Sleep Quality; DP = Depression; *p < 0.05 **p < 0.01, ***p < 0.001. Regression weights for cross-lagged SEM model. SO = Smartphone Overdependence; SQ = Sleep Quality; DP = Depression; *p < 0.05 **p < 0.01, ***p < 0.001. Fig. 1Autoregressive cross-lagged model depicting the relationships among smartphone overdependence, sleep quality, and depression. This figure illustrates the autoregressive and cross-lagged relationships among smartphone overdependence, sleep quality, and depressive symptoms across three time points. T1 = Time 1, T2 = Time 2, T3 = Time 3. All estimates are standardized regression coefficients. Asterisks (*) indicate statistically significant paths: *p < .05; **p < .01; ***p < .001.. Autoregressive cross-lagged model depicting the relationships among smartphone overdependence, sleep quality, and depression. This figure illustrates the autoregressive and cross-lagged relationships among smartphone overdependence, sleep quality, and depressive symptoms across three time points. T1 = Time 1, T2 = Time 2, T3 = Time 3. All estimates are standardized regression coefficients. Asterisks (*) indicate statistically significant paths: *p < .05; **p < .01; ***p < .001.. Significant cross-lagged effects were also observed. Perceived Sleep quality at an earlier time point negatively predicted both subsequent smartphone overdependence and depressive symptoms respectively (β = -0.041, p < .01; β = -0.042, p < .01; β = -0.036, p < .05; β = -0.035, p < .05). Additionally, depressive symptoms negatively predicted subsequent perceived sleep quality (β = -0.06, p < .001; β = -0.056, p < .001). In contrast, smartphone overdependence did not significantly predict subsequent sleep quality or depression, and depression did not significantly predict subsequent smartphone overdependence. Taken together, these findings suggest that, despite the three-year interval, perceived sleep quality had a predictive influence on both smartphone overdependence and depression, while depression in turn predicted changes in sleep quality. These results provide evidence of cross-lagged associations among the variables, highlighting perceived sleep quality as a key factor in adolescents’ mental and behavioral health. Discussion This study examined reciprocal relationships between smartphone overdependence, perceived sleep quality, and depressive symptoms over time using cross-lagged modeling. The analysis revealed that smartphone overdependence did not significantly predict later sleep quality, whereas poorer sleep quality significantly predicted subsequent smartphone overdependence, suggesting it may be a prospective predictor of greater smartphone overdependence during adolescence. Furthermore, the results provided additional empirical support for the bidirectional association between perceived sleep quality and depression, suggesting a reinforcing cycle between the two. These findings expand the current understanding of the temporal dynamics of these health-related factors and offer important insights for developing clinical interventions and policy strategies aimed at alleviating associated health problems in adolescents. This study found that perceived sleep quality significantly predicted subsequent smartphone overdependence in adolescents, indicating that poorer sleep quality is associated with higher levels of smartphone dependence over time. Although a significant concurrent association was observed, reciprocal paths were not significant in the final model, differing from prior studies reporting detrimental effects of problematic smartphone use on adolescents’ sleep28,29. Discrepancies in findings may reflect differences in how sleep was measured. Previous studies examining sleep quality and smartphone overuse also used self-report questionnaires, but incorporated multiple sleep dimensions, unlike our single-item measure. This methodological difference may partly explain inconsistent findings. Sleep quality was assessed with a single self-report item, which facilitates large-scale data collection but limits assessment of multidimensional features-onset latency, maintenance, duration, and efficiency-and precludes identification of objective sleep disturbances. Subjective sleep evaluations may reflect pubertal stage and psychosocial influences51. Given that adolescents’ sleep is shaped by both developmental and environmental factors, subjective sleep quality should be viewed as complementary to objective metrics, capturing aspects inaccessible to objective assessment and serving as a unique marker of psychological and affective adjustment during puberty52. Accordingly, while reliance on a single-item measure warrants cautious interpretation, it provides developmentally meaningful insight into perceived sleep disturbance in real-world adolescence. Another plausible explanation for this pattern is the influence of sleep on adolescents’ self-regulation and coping strategies. Sleep plays a critical role not only in adolescents’ mental health but also in their behavioral control27,53. Sleep deprivation, including insufficient sleep duration and poor sleep quality, has been shown to impair adolescents’ capacity for self-regulation, making them more susceptible to impulsive behaviors such as excessive smartphone use54. Moreover, reduced sleep quality may contribute to cognitive fatigue and emotional instability, which, in turn, lead adolescents to rely more on emotion-focused or avoidant coping strategies rather than problem-solving ones55. Such shifts in coping may encourage the use of smartphones for emotional regulation, ultimately increasing dependence on these devices56. Although no significant lagged effect of smartphone overdependence on sleep quality was observed in this study, previous research has shown that nighttime smartphone use can reduce melatonin secretion due to blue light exposure, disrupting the circadian rhythm and impairing sleep quality57. Considering these findings along with the significant correlations observed in our analysis, our results suggest a complex underlying relationship between sleep problems and smartphone overdependence. Furthermore, our study supports the predictive role of perceived sleep quality in subsequent smartphone overdependence. These findings suggest that efforts to reduce smartphone overdependence should extend beyond direct interventions targeting smartphone use. Implementing developmentally appropriate and feasible strategies aimed at improving sleep quality is essential. Specifically, interventions should address academic demands that delay sleep (e.g., self-regulated scheduling) and combine delivery methods such as in-person sessions, mobile applications, and virtual platforms to enhance engagement and sustainability58. Additionally, fostering autonomy through self-monitoring strategies, such as individualized sleep goal setting and the use of sleep diaries, may help adolescents take an active role in improving their sleep behaviors. A reciprocal relationship between sleep quality and depressive symptoms was also observed over time. Specifically, depressive symptoms at earlier time points predicted poorer sleep quality at later time points, while lower sleep quality predicted increased levels of depression. This bidirectional association aligns with previous research supporting mutual causal influences between sleep and depression in adolescents. Hill’s22 large-scale longitudinal study demonstrated that sleep disturbances were stronger predictors of adolescent mental health problems, such as depression and anxiety, than childhood adversity or family history. The study also found that depressive symptoms contributed to worsening sleep quality over time. Similarly, a review by Chai and Bian23 reported bidirectional associations between various sleep problems (e.g., reduced sleep duration, daytime sleepiness) and depressive symptoms during adolescence. These findings suggest that interventions targeting sleep should not be limited to improving sleep behavior. Rather, sleep interventions can serve as a critical pathway for enhancing emotional well-being and alleviating depressive symptoms in adolescents. Future intervention strategies should adopt integrated approaches that consider the relationship between sleep quality and depression. Improving sleep may be a viable preventive and therapeutic strategy for reducing depressive symptoms. Additionally, to ensure the sustainability and effectiveness of sleep interventions, components of emotional regulation that address psychological factors such as depression should be incorporated59. Such an approach may enhance long-term behavioral changes and maximize intervention outcomes. Limitations This study has several limitations. First, the findings were based on data from Korean adolescents, which may limit the generalizability of the results to youth populations in different racial, cultural, and socioenvironmental contexts. Future studies should examine the applicability of these findings across diverse cultural settings. Nevertheless, the study offers meaningful insights by using nationally representative longitudinal data collected over three waves to explore the temporal relationships among the key variables. Second, the primary variables-smartphone overdependence, perceived sleep quality, and depressive symptoms-were assessed through adolescents’ self-reports, which may be subject to response bias and may not fully align with objective or observer-based assessments. Furthermore, using a single-item measure of subjective sleep quality is a key limitation, as it does not capture the full complexity of sleep, including onset, maintenance, duration, and efficiency. This measurement limitation may have influenced observed associations, particularly the significant cross-lagged effect from perceived sleep quality to smartphone overdependence but not vice versa. Such asymmetry may reflect constraints of relying on a single subjective indicator rather than objective or multidimensional assessments. Future research should employ validated multi-item instruments to improve reliability and deepen understanding of these dynamic associations. Conclusion This study examined the reciprocal associations among smartphone overdependence, sleep quality, and depressive symptoms in adolescents using a cross-lagged model and nationally representative longitudinal data. A significant concurrent association was observed between smartphone overdependence and poor sleep quality. However, cross-lagged analysis revealed that poor sleep quality prospectively predicted increased smartphone overdependence. Additionally, bidirectional relationship between perceived sleep quality and depression was confirmed. These results suggest that integrated interventional strategies aimed at improving perceived sleep quality could serve as a central pathway to simultaneously reduce both smartphone overdependence and depressive symptoms. Multidimensional approaches, such as combining motivational enhancement with sleep hygiene education tailored to adolescents’ developmental needs, may promote more sustainable outcomes. Interventions targeting smartphone use should also consider sleep-related factors as part of a broader, integrated strategy. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1