Impaired sleep quality mediates the relationship between internet gaming disorder and conduct problems among adolescents: a three-wave longitudinal study

Internet Gaming Disorder (IGD) has emerged as a growing concern among adolescents [11]. IGD is characterized by excessive and compulsive gaming behaviors that result in significant impairment or distress. It is associated with various negative outcomes, including social isolation, impaired sleep quality, depression, anxiety, academic failure, and suicidality [12–14]. Emerging research has identified a positive relationship between IGD and conduct problems [15–17]. However, longitudinal studies examining this association are limited and have produced mixed results [18–21]. While some research suggests that IGD can predict future conduct problems [19], other studies have not found a significant long-term relationship [18]. These inconsistent findings highlight the need for large-scale longitudinal studies to clarify the temporal relationship between IGD and conduct problems.

A key gap in the current literature is the limited understanding of the mechanisms linking IGD to conduct problems. Most existing studies have focused on the direct relationship without exploring potential mediating factors. Impaired sleep quality is a promising candidate for mediation, which is a common consequence of behavioral addictions including IGD, social media addiction, and smartphone addiction [22, 23]. Longitudinal research indicates that IGD often leads to sleep problems, such as reduced sleep duration and poorer sleep quality [12]. These sleep issues can, in turn, result in increased irritability, impaired emotional regulation, and heightened aggression, all of which contribute to conduct problems [24, 25]. Preliminary support for this mediation comes from cross-sectional studies. For example, a large-scale American study found that increased screen time was associated with aggressive and rule-breaking behaviors, with shorter sleep duration mediating this relationship [26]. Another study reported that insomnia fully mediated the relationship between IGD and aggression in youths aged 18 to 24 years [27]. However, these studies are limited by their cross-sectional design, which does not allow for establishing temporal order. To date, no longitudinal research has rigorously tested whether impaired sleep quality mediates the relationship between IGD and conduct problems.

Both IGD and conduct problems are more prevalent among boys, and recent evidence suggests that their relationship may differ by sex. For instance, one study found a reciprocal relationship between IGD and aggression in boys but not in girls [21], highlighting the importance of considering sex differences in future research. Additionally, developmental stage may moderate the association between IGD and mental health outcomes. Research indicates that the impact of IGD on mental health may vary between early and late adolescence [28–30]. A recent study showed that the relationship between IGD, depression, and anxiety differed between middle (junior high) and late (senior high) adolescents [28]. Furthermore, a meta-analysis reported that the link between IGD and aggression is stronger in younger adolescents compared to older youths, such as college students [17].These findings emphasize the need to examine how sex and developmental stage influence the relationship between IGD, conduct problems, and impaired sleep quality.

To bridge these gaps, we conducted the present three-wave longitudinal study among a substantial sample of Chinese adolescents (N = 20137), aiming to test the following hypotheses: (1) Is IGD an independent risk factor for conduct problems among adolescents? (2) Does impaired sleep quality mediate the relationship between IGD and conduct problems? (3) Is there a difference in the association between IGD, conduct problems, and impaired sleep quality across sex and developmental stages? These subgroup analyses were explanatory in nature.

This longitudinal, school-based study was conducted in Zigong City, located in southwestern China. Detailed information on the sampling process, data collection methods, and quality assurance procedures has been previously published [31–33]. In November 2020 (Wave 1, T1), we recruited a total of 13,361 seventh-grade and 6,776 tenth-grade students from 76 schools using a two-step cluster sampling approach. Initially, all eligible middle school principals in the selected area were invited to participate. Subsequently, students from each participating school were enrolled. These participants were followed up in November 2021 (Wave 2, T2) and November 2022 (Wave 3, T3). Data were collected through electronic questionnaires, which students completed in their schools’ computer centers during regular school hours.

The study was conducted in accordance with the Declaration of Helsinki and received ethical approval from the Zigong Mental Health Center (Approval No. 2020-8-01). Written informed consent was obtained from all participants, and for those under 18, consent was also obtained from a parent or legal guardian.

Descriptive statistics were calculated for all variables, with means and standard deviations presented for continuous variables and frequencies and percentages for categorical variables. Baseline characteristics were compared between adolescents with and without conduct problems using Chi-square tests for categorical variables and Student’s t-tests for continuous variables.

Logistic regression models were then employed to examine both cross-sectional and longitudinal associations between baseline IGD and conduct problems across the three time points. These models controlled for baseline demographic factors and mental health issues. In the longitudinal analysis, we additionally adjusted for baseline conduct problems, which ensured that the observed association between IGD, impaired sleep quality, and later conduct problems reflected their unique contribution, rather than pre-existing behavioral tendencies.

Subsequently, a mediation analysis was conducted with T1 IGDS9-SF scores as the predictor, T2 PSQI total scores as the mediator, and T3 SDQ-CP scores as the outcome for participants with at least one follow-up wave. Missing data were handled using full information maximum likelihood, which estimated parameters using all available data and enhanced statistical efficiency without the need for data imputation [47]. Path coefficients were defined as follows: “a” for the effect of T1 IGD on T2 impaired sleep quality, “b” for the effect of T2 impaired sleep quality on T3 conduct problems, and “c” for the direct effect of T1 IGD on T3 conduct problems. The indirect effect was calculated as the product of paths (a * b). A bootstrap procedure with 5,000 iterations was used to generate 95% confidence intervals for both direct and indirect effects, with effects considered significant if the confidence intervals did not include zero. Both unadjusted and adjusted models were tested. As an additional robustness check, we reran the mediation analysis using only participants with complete data across all waves.

Finally, multiple-group analyses were performed to explore differences by gender (boys vs. girls) and developmental stage (seventh-grade as early adolescents vs. tenth-grade as late adolescents). Chi-square difference tests were used to compare the fit of an unconstrained model against models with paths constrained to be equal across the two groups.

All descriptive analyses, group comparisons, and logistic regressions were conducted using R (version 4.20). Mediation analyses were performed using Mplus (version 8.3). Statistical significance was set at p < 0.05 for all two-tailed tests.

Table S1 presents the baseline characteristics of participants across all three waves. Initially, 20,137 seventh and tenth-grade students were enrolled in the study. At the first follow-up (T2), 15,061 students participated (response rate: 75%), and at the second follow-up (T3), 14,706 students completed the survey (response rate: 73%). Approximately 85% of the adolescents participated in at least one follow-up (n = 16,982), and 63% (n = 12,785) participated in all three waves. There were no significant differences in the main study variables—IGD status, conduct problems, and sleep quality—between participants who remained in the study and those who dropped out. However, dropout was slightly more likely among girls, younger students, only children, those with parents of higher education, and seventh-grade students, although these differences were minimal.

At baseline, the prevalence of conduct problems was 9.0% (Table 1). Adolescents with conduct problems were significantly younger (mean age = 13.24 ± 1.36 years) compared to those without (mean age = 13.42 ± 1.46 years). A higher proportion of boys (53.0% vs. 49.1%) and seventh-grade students (73.7% vs. 65.6%) exhibited conduct problems compared to their counterparts without conduct problems (p < 0.05). Additionally, conduct problems were more prevalent among only children (28.0% vs. 22.0%), those from single-parent or remarried families (26.5% vs. 20.1%), and those engaged in substance use (alcohol: 31.6% vs. 15.4%; smoking: 18.5% vs. 5.8%, p < 0.001). Psychologically, adolescents with conduct problems reported more severe impaired sleep quality (PSQI: 6.0 ± 3.8 vs. 3.8 ± 3.0) and higher IGD scores (IGDS9-SF: 19.8 ± 8.7 vs. 14.5 ± 6.0, p < 0.001). Furthermore, the prevalence of emotional problems, peer problems, and hyperactivity/inattention was significantly higher among those with conduct problems.

The prevalence of conduct problems decreased over time, from 9.0% at T1 to 6.5% at T2 and 6.0% at T3. As illustrated in Fig. 1, both risky gamers and gamers with IGD consistently reported higher prevalence rates of conduct problems compared to healthy gamers across all three time points. Specifically, at baseline, 34.7% of gamers with IGD and 19.3% of risky gamers exhibited conduct problems, compared to 6.3% of healthy gamers. Similar trends were observed at T2 and T3.

Logistic regression analyses revealed that both cross-sectional and longitudinal associations between IGD and conduct problems remained significant after adjusting for baseline covariates (Table 2). Compared to adolescents without problematic gaming, risky gamers had increased odds of conduct problems at T1 (adjusted odds ratio [OR] = 1.41, 95% CI = 1.13–1.77, p = 0.003), T2 (adjusted OR = 1.33, 95% CI = 1.12–1.58, p = 0.001), and T3 (adjusted OR = 1.32, 95% CI = 1.10–1.58, p = 0.003). IGD gamers exhibited even higher odds at T1 (adjusted OR = 3.01, 95% CI = 2.41–3.76, p < 0.001), T2 (adjusted OR = 1.63, 95% CI = 1.20–2.22, p = 0.002), and T3 (adjusted OR = 1.52, 95% CI = 1.09–2.11, p = 0.013).

Additionally, baseline impaired sleep quality was significantly associated with increased odds of conduct problems at all three time points: T1 (adjusted OR = 1.40, 95% CI = 1.24–1.58, p < 0.001), T2 (adjusted OR = 1.49, 95% CI = 1.28–1.74, p < 0.001), and T3 (adjusted OR = 1.48, 95% CI = 1.26–1.74, p < 0.001).

The mediation analysis revealed both direct and indirect effects of IGD on conduct problems (Fig. 2). In the unadjusted model, IGD at T1 was significantly associated with impaired sleep quality at T2 (a = 0.218, 95% CI = 0.201–0.235), which in turn were associated with increased conduct problems at T3 (b = 0.233, 95% CI = 0.214–0.251). The direct effect of IGD on conduct problems (c) was 0.128 (95% CI = 0.110–0.146), and the indirect effect (a × b) was 0.051 (95% CI = 0.045–0.056). This indicates that impaired sleep quality mediated 28.5% of the total effect in the unadjusted model. After adjusting for baseline covariates, all paths remained significant. The direct effect was reduced to 0.043 (95% CI = 0.022–0.063), and the indirect effect decreased to 0.009 (95% CI = 0.005–0.013), accounting for 17.3% of the total effect.

Sensitivity analysis using data from participants with complete data across all waves produced consistent results. In the fully adjusted model, the direct effect of IGD on conduct problems was 0.041 (95% CI = 0.020–0.062), and the indirect effect via impaired sleep quality was also significant (β = 0.010, 95% CI = 0.006–0.014). These findings further support the reliability of our mediation model.

Figure 3 displayed the results of subgroup analysis. Multiple-group analyses revealed significant differences in the mediation model between boys and girls (Δχ² = 101.266, Δdf = 33, p < 0.001). While in both girls and boys, the direct and indirect effect of IGD on conduct problems were significant, both the total effect and direct effect was greater in boys (total effect = 0.061, direct effect = 0.051) compared to girls (total effect = 0.037, direct effect = 0.029). The mediation proportion of impaired sleep quality was slightly higher among girls (21.6%) compared to boys (16.4%).

There were also significant differences in the mediation model between early and late adolescents (Δχ² = 137.718, Δdf = 33, p < 0.001). Both early (seventh-grade) and late (tenth-grade) adolescents showed significant direct and indirect effects of IGD on conduct problems, with comparable total effects. However, the proportion of the mediation effect via impaired sleep quality was higher among early adolescents (23.5%) compared to late adolescents (10%). This indicates that impaired sleep quality played a more substantial mediating role in the IGD-conduct problems relationship during early adolescence.

At baseline, the prevalence of conduct problems and IGD was 9.0% and 2.6%, respectively, consistent with recent large-scale epidemiological data on Chinese adolescents [3, 48, 49]. While most prior research on the association between IGD and conduct problems has been conducted in Western populations [15, 16], our findings replicate these associations within a substantial Chinese sample. We observed a strong cross-sectional and longitudinal relationship between IGD and conduct problems. Notably, even adolescents classified as risky gamers, who did not meet the criteria for IGD, exhibited a higher risk of concurrent and future conduct problems. These results suggest that even subclinical levels of problematic gaming are linked to increased behavioral issues. This cross-cultural consistency underscores the universal nature of the IGD-conduct problems relationship and highlights the importance of screening for conduct problems among adolescents engaged in problematic gaming behaviors, including both risky gamers and those with IGD.

Several explanations may account for the substantial association between IGD and conduct problems. First, excessive gaming may displace time spent on social activities [50, 51], leading to reduced social skills and increased interpersonal conflicts, such as family disputes and peer bullying [52], which can contribute to conduct problems. Second, adolescents with IGD often struggle with difficulties in emotional regulation and expression and possess inadequate coping strategies [53–55]. Reliance on gaming for emotional regulation may hinder the development of healthier coping mechanisms, exacerbating behavioral issues [56]. Finally, from a neurobiological perspective, IGD is associated with alterations in brain regions involved in executive functioning [57–59], such as the prefrontal cortex, which is critical for impulse control and decision-making. Structural and functional changes in these areas may impair adolescents’ ability to regulate behaviors and adhere to social norms, thereby fostering conduct problems.

Our study found that impaired sleep quality is an independent predictor of conduct problems, consistent with prior research [60]. Poor sleep disrupts emotional regulation, resulting in heightened irritability, impulsivity, and aggression [24, 25].It also impairs executive functions, such as impulse control and decision-making [61], which increases the likelihood of risk-taking and rule-breaking behaviors. These effects may be underpinned by biological mechanisms, including dysregulation of the hypothalamic-pituitary-adrenal axis, alterations in the brain’s cholinergic system, and structural changes in key brain regions [62–64]. Together, these pathways highlight the critical role of sleep disturbances in the development and exacerbation of conduct problems.

Consistent with our hypothesis, impaired sleep quality significantly mediated the relationship between IGD and conduct problems, accounting for about one-fifth of the total effect after adjusting for covariates. This finding contributes to the growing body of literature recognizing impaired sleep quality as a crucial pathway linking IGD to a series of adverse mental health outcomes, including depression, suicidality, psychotic-like experiences, and impaired well-being [65–69]. Collectively, these results highlight the importance of incorporating sleep assessments and interventions into strategies aimed at mitigating the behavioral and psychological impacts of IGD. Comprehensive intervention programs that target both gaming behaviors and sleep health could more effectively support the well-being of affected adolescents.

Our study revealed significant sex differences in the IGD-sleep-conduct problems relationship. Specifically, both the direct and total effects of IGD on conduct problems were more pronounced among boys, indicating that IGD may be a stronger predictor of conduct problems in this group. These findings partially align with those of Kim et al. [21], who observed a longitudinal association between IGD and aggression in Korean adolescent boys but not in girls. One possible explanation for these sex differences is the type of games played; boys may engage more frequently in aggressive and violent gaming, which could potentially lead to higher conduct problems [70]. Additionally, previous research on IGD and depression has produced similar results, finding a stronger effect of IGD on later depression among boy [71, 72], though the results were not entirely consistent [29, 73]. Collectively, these findings might suggest that boys may be more vulnerable to IGD-related mental distress. Future studies should explore the underlying biological and psychosocial mechanisms contributing to these sex differences.

Regarding developmental stages, the mediation effect of impaired sleep quality was more pronounced among early adolescents (seventh-grade students) compared to late adolescents (tenth-grade students). While the total effects of IGD on conduct problems were comparable across both groups, impaired sleep quality accounted for a larger proportion of this relationship in early adolescents. This suggests that younger adolescents may be more vulnerable to the negative impacts of IGD on sleep, which in turn affects their behavioral outcomes. Early adolescence is a critical period for developing sleep patterns and behavioral regulation; thus, interventions targeting sleep hygiene may be particularly beneficial for this age group.

Our study has several public health and clinical implications. The robust cross-sectional and longitudinal associations between IGD and conduct problems underscore the necessity for early identification and targeted interventions for adolescents engaged in problematic gaming. Recognizing IGD as a substantial risk factor for conduct problems emphasizes the importance of screening for both conditions in educational and clinical settings. Comprehensive screening programs that include assessments for IGD, conduct problems, and sleep quality can facilitate timely and effective interventions. The mediating effect of impaired sleep quality between IGD and conduct problems suggests that targeting sleep problems simultaneously with gaming behaviors may be an effective strategy for reducing gaming-related conduct problems. For example, interventions could integrate cognitive-behavioral therapy for insomnia (CBT-I) to improve sleep quality alongside CBT for IGD to address compulsive gaming. Additionally, psychoeducational programs could teach adolescents and parents about the importance of sleep hygiene and healthy screen time habits. School-based initiatives, such as sleep awareness campaigns and structured gaming schedules, could further reinforce these efforts. Future research is needed to validate the efficacy of such integrated approaches. Tailored intervention strategies are warranted based on subgroup analyses: girls and early adolescents may benefit more from interventions targeting impaired sleep quality, whereas boys and late adolescents might require more direct interventions addressing IGD itself.

The present study boasts several strengths, including a large sample size and a robust three-wave longitudinal design, which enhance the reliability and generalizability of the findings. However, several limitations must be acknowledged. First, the reliance on self-reported measures may introduce response biases, such as social desirability or recall bias, potentially affecting data accuracy. Second, the study’s focus on Chinese adolescents may limit the generalizability of the findings to other cultural or geographical populations. Third, although impaired sleep quality was examined as a mediator, other potential mediators, such as emotional regulation, family dysfunction, and cyberbullying, were not explored and may also play critical roles in the IGD-conduct problems relationship. Additionally, while our three-wave longitudinal design supports temporal inferences, the observational nature of the study limits causal conclusions, and reverse causation remains a possibility. Future research should address these limitations by incorporating objective measures, diverse populations, and additional mediating factors to further clarify the mechanisms linking IGD to conduct problems.