Informed consent was obtained from students’ parents or guardians on behalf of the minors. This study protocol was approved by the Ethics Committee of the Zhejiang Provincial Center for Disease Control and Prevention (2020-040-002).

Data for this cross-sectional study were derived from the 2021 China Common Disease and Risk Factor Surveillance Among Students conducted in Zhejiang Province, an annual school-based health survey [44,45]. A stratified cluster sampling was conducted to randomly select one urban area and one rural area from all 11 prefecture-level cities in Zhejiang Province. Schools were then randomly selected across the urban (including two primary schools, two middle schools, two high schools, one vocational high school, and one university) and rural (including two primary school, two middle school, and one high school) regions, covering students from grade 1 and above. To ensure adequate representativeness, we randomly recruited a minimum of 80 students per grade by class units [46,47]. All the students underwent physical examinations implemented by trained health care workers. Students of grade 4 and above were asked to fill out a self-administered questionnaire about their health-related behaviors, and teachers and trained research assistants were present to provide guidance to the students. The initial sample of this study included a total of 21,737 students from grade 4 and above. Furthermore, students with missing data on depressive symptoms (n=57) and covariates (n=344) were excluded, leaving a total of 21,336 students in the final analysis (Figure 1).

The questionnaire asked respondents to report their weekly internet use time and indicate whether they had exhibited specific IA behaviors in the past week. The IA behaviors were assessed using a 10-item IA questionnaire based on the revised version of Young’s diagnostic questionnaire (Cronbach α=0.778) [21,48] (Table S1 in Multimedia Appendix 1). Internet use time was categorized into four groups: 0, 0.1‐1, 1.1‐2, and >2 hours per day. The number of IA behaviors was categorized as follows: 0, 1, 2, or ≥3 [24].

Depressive symptoms were assessed using the Chinese version of the Center for Epidemiologic Studies Depression Scale, which is a 20-item scale constructed to measure the level of depressive symptoms over the past week [49]. The reliability and validity of the Center for Epidemiologic Studies Depression Scale among the Chinese population have been previously demonstrated [50-52], and the Cronbach α in our study was 0.877. Participants indicated the frequency they encountered each item in the past week on a 4-level scale [53]. Responses ranged from “rarely or none of the time (less than 1 day, 0 point)” to “most or all of the time (5‐7 days, 3 points).” The total score ranged from 0 to 60, with a score of 16 or higher indicating the presence of depressive symptoms [49,54].

Demographic information was collected using a validated questionnaire [44,46]. Anthropometric data were measured by trained health staff with standardized equipment. Weight and height were recorded to the nearest 0.1 kg and 0.1 cm, respectively, with participants removing shoes and heavy coats. Covariates adjusted in the study included age, gender (boy or girl), residence (urban or rural), ethnicity (Han or others), number of family members (≤2, 3, 4, or ≥5), grade (primary school: grades 4‐6; middle school: grades 7‐9; or high school; grades 10‐12), boarding (yes or no), BMI (calculated as kg/m²), and sedentary time (hour per day).

In the analysis, all continuous variables exhibited skewness in normality tests and were reported as medians and IQRs. Categorical variables were presented as frequencies and relative percentages (%). To assess differences in characteristics based on the presence or absence of depressive symptoms among students, the Wilcoxon rank sum test was used for continuous variables and the chi-square test was used for categorical variables.

To investigate the association of internet use time (as a categorical variable) and the number of IA behaviors (as a categorical variable) with the risk of depressive symptoms, logistic regression models were performed, and results were presented as the odds ratio (ORs) with 95% CIs. Then, the logistic regression models were stratified by sex and grade. Restricted cubic spline (RCS) curve was used to reveal the dose-response association of internet use time and the number of IA behaviors met with the risk of depressive symptoms. In addition, linear regression models were applied to assess the association of internet use time and IA behaviors with the depressive score, and the dose-response correlation was assessed with 4-knots (obtained at the quartiles of internet use time or number of IA behaviors) RCS curve. All data analyses were performed using STATA 16.0 (StataCorp LLC).

In our study, longer internet use time was associated with a higher risk of depressive symptoms among Chinese children and adolescents overall, for both sexes, and among primary and middle school students. Interestingly, in grades 10‐12, no significant association between internet use time and depressive symptoms was observed. A notable J-shaped association between internet use and depressive symptoms was observed overall among boys and students in grades 4‐6. For IA, a larger number of IA behaviors was associated with a higher risk of depressive symptoms, with a significant nonlinear association in total students, girls, and students in grades 4‐6. Moreover, all participants but high school students with longer internet use time presented higher depressive scores. For IA behaviors, a higher number was associated with a higher depressive score in both sexes and all grades.

Our findings added to the existing body of research on the impact of internet use on mental health but with some differences. Previous studies have reported a positive association between heavy internet use and depressive symptoms [55,56]. The observed J-shaped association also highlighted the possibility of a protective effect associated with a minimal amount of internet use (around 0.5 h/day). This could be explained by the positive aspects of internet use, such as social connection, access to support networks, and educational resources [57,58]. While excessive use could lead to isolation, sleep disturbances, and exposure to harmful content, thereby increasing the risk of depression [59-61]. The displacement theory indicates that heavy internet use would reduce children’s intimacy with their family members as well as social involvement, which may lead to stressful interpersonal relationships and increase psychological problems [43]. The sex-specific differences in the associations between internet use time and depressive symptoms could be due to the different ways in which boys and girls use the internet, or the difference in how they respond to online experiences. For instance, girls might be more affected by social comparisons or cyberbullying, which are common on online platforms [62,63]. For the diminished association between internet use time and depressive symptoms among senior grades, one explanation for this could be the maturation and development of coping strategies in older adolescents, enabling them to handle the stresses associated with the internet more effectively [64]. Another plausible explanation might be that older students are using the internet for more educational purposes, which may not have the same impact on depressive symptoms as recreational use [65].

Our study provides compelling evidence that a higher number of IA behaviors are significantly associated with a higher risk of depressive symptoms among students. These findings align with previous research suggesting a strong link between IA and depressive symptoms. For instance, a systematic review by Lopes et al [37] found a significant association between IA and depression. Our study adds to this body of evidence by demonstrating a nonlinear increase in the risk of depressive symptoms with an increasing number of IA behaviors, which could be due to the increasing disruption of normal life activities, social isolation, and exposure to negative online experiences with higher levels of IA [66]. The increase in depressive risks was more pronounced in the early stage (≤3) of IA behaviors, which underscored the importance of early control of excessive internet use and IA symptoms. Similar to the findings for the associations between internet use time and depressive symptoms, the observed sex differences between IA and depressive symptoms might also be due to the vulnerability of girls in the online activities, such as social comparisons or cyberbullying [67]. The varying impact of IA behaviors across different school grades could also be explained by the resilience and coping mechanisms among students in different developmental stages, where senior students might have developed better digital literacy skills and coping strategies to manage the potential negative effects of IA. To mitigate the negative impact of excessive internet use, parents should attach importance to creating a positive home media environment, as well as involving their children in alternative activities like reading and exercising [31].

There are several strengths in our study. First, we used the RCS curve to show the nonlinear dose-response association. This approach provides a better fit to the data and reduces the degrees of freedom, enhancing the accuracy, reliability, and precision of our study. Second, the database used in this study was derived from a large representative sample of children and adolescents in Zhejiang Province, China. Moreover, sex- and grade-specific subgroup analyses were performed separately, capturing the variations in the association between internet use and depressive symptoms.

However, our study is not without limitations. The cross-sectional design precluded our ability to establish causality. A lack of detailed information regarding the age of IA initiation and family environment factors, all of which could potentially influence adolescent depressive symptoms, limited the depth of our analysis. Furthermore, our participant sample was drawn exclusively from an economically developed region of China. This may limit the extent to which our findings are representative of the broader population of Chinese children and adolescents.

A significant association between internet use and depressive symptoms was demonstrated, with excessive internet use and more IA behaviors being associated with higher risks of depressive symptoms. A notable J-shaped association between internet use time and depressive symptoms was observed among boys and primary school students. Early preventive and mitigating interventions aimed at promoting appropriate internet use should be prioritized to prevent depressive symptoms among children and adolescents. Teachers and parents should help children develop healthy internet use habits and monitor children’s internet activities actively. Future studies could explore the longitudinal associations between internet use at different developmental stages and subsequent mental disorders in youth and investigate potential sex differences.