Internet gaming disorder and depression mediated by impaired resilience and sleep distress: a three-wave longitudinal study among Chinese adolescents

Digital addictions, including internet gaming disorder (IGD), smartphone addiction, internet addiction and social media addiction, are regarded as emerging risk factors for adolescent depression (Chen et al., 2020a; Düll et al., 2024; Karakose et al., 2023). Among these, IGD – characterized by excessive and problematic gaming that causes distress and functional impairment – has been one of the most extensively studied (Düll et al., 2024). Numerous cross-sectional studies have consistently shown a positive association between IGD and depression, with a moderate effect size of 0.33 (Alimoradi et al., 2024; Richard et al., 2020; Yoon et al., 2021). Meta-analyses have also revealed a high comorbidity rate of depressive disorder among individuals with IGD, approximately 34% (Ostinelli et al., 2021).

While cross-sectional studies have established a strong link between IGD and depression, the question of whether IGD predicts future depressive symptoms remains debated. A recent systematic review of cohort studies revealed mixed results: six studies supported IGD as a predictor of depressive symptoms, while seven found no such relationship (Düll et al., 2024). Longitudinal studies in Chinese adolescents have shown similar inconsistencies: several studies during the COVID-19 pandemic found that IGD was associated with worsening depressive symptoms (Chen et al., 2022a, 2022b; Chen et al., 2021), while other studies suggested that IGD may be an outcome, rather than a predictor, of depression among adolescents (Chang et al., 2022a; Teng et al., 2021; Yin et al., 2023). Moreover, few studies have explored the mechanisms through which IGD may lead to depression, highlighting the need for large-scale longitudinal research to clarify this association and identify key mediating factors.

Resilience, a well-known protective factor against mental distress (Watters et al., 2023), may serve as a crucial mediator between IGD and depression. Although longitudinal evidence on the relationship between resilience and IGD is limited, cross-sectional studies consistently observed impaired resilience in individuals with IGD (Canale et al., 2019; Liu et al., 2024; Ma et al., 2024). Additionally, longitudinal research on other behavioural addictions, such as mobile phone and internet addiction, has shown that these addictions can weaken resilience (Li et al., 2022; Shang et al., 2024; Yao et al., 2024). These studies also confirm that reduced resilience mediates the relationship between these addictions and mental health issues, including depression and eating disorders (Li et al., 2022; Shang et al., 2024). Given this evidence, it is plausible that IGD similarly diminishes resilience, making adolescents more vulnerable to depression by weakening their ability to cope with the emotional and psychological challenges. However, no prior studies have specifically tested the mediating role of resilience in the relationship between IGD and depression.

Sleep distress is another potential mediator linking IGD to depression. Longitudinal studies have consistently reported the predictive role of digital addictions on subsequent sleep distress (Byeon et al., 2022; Chang et al., 2022b; Saffari et al., 2023). In addition, recent research suggests that sleep distress may be a common pathway linking those addictions to mental health problems. For example, Li et al. found that insomnia mediated 60.6% of the relationship between internet addiction and depression and 44.8% of the relationship between social media addiction and depression among adolescents (Li et al., 2017). Similarly, sleep distress has also been shown to mediate the relationship between smartphone addiction and depression in college students (Liu and Lu, 2022; Zou et al., 2019). In the context of IGD, studies have consistently reported the mediating role of sleep problems between IGD and various negative outcomes, such as psychotic experiences, violence and suicidality (Bersani et al., 2022; Fekih-Romdhane et al., 2023; Garg et al., 2023; Liao et al., 2024). However, only two studies have investigated the relationship between IGD, depression and sleep distress, and both were cross-sectional (Alshammari et al., 2022; Li et al., 2019). More longitudinal research is needed to confirm this mediation effect.

Previous longitudinal studies have demonstrated a link between sleep distress and resilience, indicating that sleep problems may lead to impaired resilience over time (Li et al., 2024; Yao et al., 2024). Consequently, it is plausible that the relationship between IGD and depression could be sequentially mediated, first through sleep distress and subsequently through resilience. In addition to the single mediating roles of each factor, this potential chain mediation may also impact the IGD-depression association.

To address these gaps, the present study was conducted to test the following hypotheses: (1) IGD predicts future depression in adolescents, and (2) there is a serial multiple mediation effect of sleep distress and resilience on the association between IGD and depression. Given the known sex differences in both IGD and depression (Marraudino et al., 2022), we also conducted exploratory subgroup analyses based on gender.

This three-wave, school-based longitudinal study was conducted in Zigong, a city in southwest Sichuan, China. The baseline assessment took place in November 2021, involving 44,270 students from grades 7 and 10 across 166 middle schools, recruited using a cluster sampling method. After excluding invalid responses due to excessively short response times, logical inconsistencies (e.g., abnormal age) and incorrect identification information (e.g., erroneous student IDs), the final baseline sample included 41,215 students, reflecting a response rate of 93.1%. Participants were re-assessed in November 2022 (T2) and November 2023 (T3).

At each survey, participants completed the electronic survey during class hours in the computer centre of their respective schools. Before the study commenced, head teachers and psychological teachers received training on the study procedures and measurement tools. They introduced the study’s purpose, supervised the survey administration and answered any questions from the participants.

Participants were informed that their involvement in the study was entirely voluntary, with the option to withdraw at any point. Written informed consent was obtained from all participants, and for those under 18, consent was also secured from their parents. The study protocol was reviewed and approved by the Ethics Committee of the Zigong Mental Health Center (Approval No. 2021003).

Continuous variables were summarized using means and standard deviations, while categorical variables were presented as frequencies and percentages. Baseline characteristics between adolescents with and without depression were compared using Chi-square tests and Student’s t-tests. To evaluate the association between baseline IGD and depression at both baseline and follow-up, univariate and multivariate logistic regression analyses were conducted.

Mediation analysis used T1 IGDS9-SF scores as predictors, T2 PSQI and CD-RISC-10 scores as serial mediators and T3 PHQ-9 scores as outcomes. Missing data were handled using full information maximum likelihood methods. This approach utilizes all available information in the dataset, even when some variables are missing for certain cases, and is known to produce less biased and more efficient estimates compared to listwise deletion or simpler imputation methods (Niu, 2024). However, this method does not include cases where participants did not complete any follow-up assessments. Baseline demographics, anxiety, sleep distress, depression and resilience were controlled as covariates. Bootstrapping (5000 samples) was used to estimate 95% confidence intervals (CIs) for the path coefficients.

We conducted a series of sensitivity analyses to assess the robustness of our findings: (1) re-analysing the data by including only participants with complete data across all three waves (n = 28,916); (2) re-analysing the data by including only baseline gamers (n = 28,558); (3) re-analysing the model by treating IGD, sleep problems, resilience and depression as a binary variable and (4) removing the sleep-related item from the PHQ-9 to minimize shared variance between the PHQ-9 and PSQI.

Given the well-established sex difference in IGD and depression (Marraudino et al., 2022), we also performed subgroup analysis using multiple-group analysis. Chi-square difference tests were performed to compare an unconstrained model with models where the main paths (IGD → Resilience, IGD → Sleep Distress, Resilience → Depression, Sleep Distress → Depression, Sleep Distress → Resilience and IGD → Depression) were constrained to be equal across sex. Baseline covariates were controlled in both models.

Descriptive analysis, group comparisons and logistic regression were performed using R (version 4.20). Mediation analysis was conducted based on path analysis using Mplus (version 8.1). All mediation models in our study (both unadjusted and adjusted) were saturated. Consequently, model fit indices were not reported. Statistical significance was set at p < 0.05 for all two-tailed tests.

The indirect effects of IGD on depression were observed through resilience (β = 0.025, 95% CI = 0.022–0.027), through sleep distress (β = 0.101, 95% CI = 0.095–0.107) and through the sequential pathway of sleep distress followed by resilience (β = 0.021, 95% CI = 0.019–0.023). The total effect of IGD on depression was 0.242 (95% CI = 0.229–0.254), with indirect effects accounting for 60.7% of the total effect – 10.3% through resilience, 41.7% through sleep distress and 8.7% through the sequential pathway of sleep distress to resilience. All paths remained significant after adjusting for baseline characteristics, with a total mediation ratio of 33.3%.

Sensitivity analyses were conducted to evaluate the robustness of our findings. First, mediation analyses using participants with complete data across all three waves (n = 28,916) were consistent with the primary analysis, with mediation ratios of 59.1% in the unadjusted model and 29.8% in the fully adjusted model (Table S1). Analyses focusing on gamers (n = 28,558) also confirmed the serial mediating roles of sleep distress and resilience, with mediation ratios of 61.3% (unadjusted) and 35.4% (adjusted) (Table S2). Treating variables as categorical yielded partial alignment with the main analysis: in the unadjusted model, indirect effects through sleep distress, resilience and their sequential pathway accounted for 64.9% of the total effect, whereas in the adjusted model, only sleep distress remained significant (mediation ratio: 18.8%) (Table S3). Finally, excluding the sleep-related item from the PHQ-9 to reduce shared variance with the PSQI yielded consistent results, with mediation ratios of 59.1% (unadjusted) and 32.7% (adjusted) (Table S4). Collectively, these analyses support the robustness of our findings.

To our knowledge, this is the largest cohort study to date investigating the predictive effect of IGD on depressive symptoms. Our findings establish IGD as a significant risk factor for depressive symptoms among adolescents, underscoring the importance of early identification and intervention strategies targeting IGD to prevent depression in this vulnerable population.

Several mechanisms may explain the direct effect of IGD on depression. First, displacement theory suggests that the immersive and competitive nature of online gaming reduces time spent on social interactions, leading to social isolation and diminished social support, both of which increase the risk of depression (Fong et al., 2024; Krossbakken et al., 2018). Second, the intense focus on gaming can result in academic neglect and poor school performance (Hawi et al., 2018), which in turn generates stress from family or teacher conflicts and feelings of inadequacy, further contributing to depression. Third, from a neurobiological perspective, IGD and depression may share similar brain function impairments. For instance, IGD is associated with dysregulation in the brain’s reward system (Weinstein and Lejoyeux, 2020), including heightened sensitivity to gaming-related rewards and diminished responsiveness to other positive stimuli. This imbalance may reinforce gaming as a primary coping mechanism while reducing the ability to derive satisfaction from other life activities, thereby contributing to depression. Structural and functional changes in the prefrontal cortex, a region critical for impulse control, decision-making and emotional regulation, have also been observed in individuals with IGD (Yan et al., 2021). These alterations can impair an adolescent’s ability to manage negative emotions and stress, increasing their vulnerability to depressive symptoms. Choi et al. provided direct evidence that lower grey matter density in the left dorsolateral prefrontal cortex mediated the relationship between gaming and depressed mood (Choi et al., 2017). However, much of the current neuroimaging evidence is based on cross-sectional data; therefore, future longitudinal studies are needed to further explore these mechanisms.

Our study is also the first longitudinal investigation into the association between IGD and resilience. In line with findings by Shang et al. in the field of internet addiction (Shang et al., 2024), our study demonstrated that resilience serves as a mediator between IGD and depression. This finding suggests that IGD may weaken an adolescent’s resilience, thereby increasing their vulnerability to depression. Individuals with IGD often use gaming as a maladaptive coping strategy, avoiding real-life problems and preventing the development of healthier ways to manage stress (Brand et al., 2016, 2019; King and Delfabbro, 2018). This avoidance behaviour, combined with the emotional instability and cognitive impairments caused by excessive gaming – such as preoccupation with gaming, difficulties in decision-making and self-regulation – further erodes their capacity to adapt to challenges (King and Delfabbro, 2018). The erosion of resilience, in turn, may make adolescents more vulnerable to developing depressive symptoms, as they become less capable of managing the emotional and psychological consequences of their gaming behaviours and other life stressors.

Consistent with previous cross-sectional studies (Alshammari et al., 2022; Li et al., 2019), we found that sleep distress is a key mediator between IGD and depression, accounting for approximately half of the total effects. IGD may contribute to sleep problems through several interconnected pathways, including difficulties in falling asleep, reduced sleep duration and impaired melatonin production due to the blue light emitted by digital devices (Kristensen et al., 2021; Lemola et al., 2015). Importantly, emerging studies suggest a bidirectional relationship between sleep problems and gaming (Liu et al., 2021, 2023; Poulain et al., 2019), indicating a potential vice cycle between the two conditions. Furthermore, sleep problems have been shown to play a crucial mediating role in the relationship between IGD and its various negative psychological consequences (Bersani et al., 2022; Fekih-Romdhane et al., 2023; Garg et al., 2023; Liao et al., 2024). These findings underscore the importance of addressing sleep disturbances in adolescents with IGD to break the cycle of IGD, poor sleep and worsening mental health.

Consistent with Hypothesis 2, we identified a significant indirect effect of IGD on depression through the sequential pathway of sleep distress followed by resilience impairment. This finding suggests that IGD may initially contribute to sleep distress, which subsequently impairs resilience and ultimately leads to depression. This result aligns with prior studies that have identified impaired resilience as a critical mediator between sleep problems and various mental health outcomes, including anxiety symptoms, psychological distress and suicidality (Cai et al., 2023; Jiang et al., 2024; Li et al., 2024). Thus, regular assessment and strengthening of resilience may help mitigate the impact of sleep disturbances on depression.

Our study revealed significant sex differences in the mediating effects of resilience and sleep distress. Despite extensive research on gender differences in IGD and depression, few studies have examined gender differences in the relationship between these conditions and their mediators. Although the exact mechanisms remain unclear, our results suggest that targeting sleep problems and resilience might be particularly effective in boys with IGD to reduce IGD-related depression.

Our study offers several key clinical implications. First, the direct and indirect effects of IGD on depression highlight the need for early identification and intervention for IGD to reduce the risk of depression. It is crucial for parents, educators and mental health professionals to be vigilant about the heightened risk of depression linked to IGD and to actively monitor and manage gaming behaviours in adolescents. Second, with one-third of IGD’s impact on depression mediated by sleep distress and impaired resilience, enhancing resilience and improving sleep quality should important intervention strategies. Resilience-building programs, such as cognitive-behavioural therapy focused on coping skills and stress management, combined with promoting healthy sleep practices – like limiting gaming before bed – may be particularly effective in mitigating the effects of IGD on depression (Joyce et al., 2018; Llistosella et al., 2023).

While our study has several strengths, including a large sample size, unbiased cluster sampling across multiple centres and a longitudinal design, it is not without limitations. First, the reliance on self-reported data for IGD, depression, resilience and sleep distress may introduce biases, such as social desirability and recall bias. Despite the high accuracy and validity of the PHQ-9 and IGDS9-SF, both scales are designed for screening rather than diagnosing depression and IGD. Future studies with clinical interviews are warranted to further validate our findings. Second, although various baseline covariates were controlled for, key factors such as academic stress, childhood trauma and other adverse experiences were not measured, which may have introduced residual confounding. Future studies should include these factors. Third, the generalizability of our findings may be limited, as the study was conducted in a single city in China, and all participants were middle school students. Future research should aim to replicate these findings in adolescents from diverse cultural, educational and clinical settings.