The current study was conducted as a part of the Surveillance of Students’ Common Diseases and Health Influencing Factors (SSCDHIF) in Shanghai, China, during September to November from 2020 to 2022.

A multi-stage stratified cluster sampling method was employed to create a representative sample of middle school students in Shanghai. From each of the 16 districts, two junior high schools, two general senior high schools, and one vocational high school were randomly selected. Within each selected school, two to three classes per grade (grades 1–3) were randomly chosen, and all students in these classes completed an annual questionnaire survey. 19687 students participated in the survey in 2020, and 19565 participants had valid data. 21669 students participated in the survey in 2021, and 19875 participants had valid data. 20568 students participated in the survey in 2022, and 18680 participants had valid data. After matching with a unique identification code, a total of 3659 students who participated in the surveillance program for 3 consecutive years (2020–2022) were included in this study ( Figure 1 ). These students progressed from first grade in 2020 to third grade in 2022. All participants and their parents/guardians have obtained informed consent before participation. The study protocol received ethical approval from the Shanghai Municipal Center for Disease Control and Prevention Ethics Committee (Approval No. 2022-13).

All the analyses were performed using R Language (V4.4.1) and SPSS 29.0. A two-tailed p value of < 0.05 was considered statistically significant. Quantitative variables were analyzed as means (standard deviation) according to the normality of distribution, and qualitative variables were analyzed as numbers (percentages). The t-test and chi-squared test were used to compare difference between sub-groups. The analytic data included depressive symptoms, using screen in the dark and other variables in each assessed interval from 2020 to 2022, and was converted to long data before Generalized Estimating Equation (GEE) analysis. Associations between using screen in the dark, screen time, sleep duration and depressive symptoms were assessed using GEE for repeated measures analysis. Model 1 was crude model. Model 2 was adjusted for district, gender, age. Model 3 was further adjusted for one-child status, family economic status, drinking history, academic performance, school bullying. The joint effect between using screen in the dark and screen time or sleep duration was also assessed.

A total of 3659 participants were included in the analysis. Their age ranged from 11 to 19 years, and the mean age was 14.41(SD =1.55) years, 51.87% (1898) were boys and 48.13% (1761) were girls. The average score of CES-D was 11.06 (SD = 8.7). Adolescents who used screen in the dark exhibited higher CES-D scores compared to those who did not. A total of 453 participants (12.38%) reported using screens in the dark. They were more likely to have drinking history, poor family economic status, poor academic performance, insufficient sleep and screen time >2 hours/day. The demographic and behavior characteristics of adolescents at baseline are shown in Table 1 .

Table 2 shows nearly a quarter of adolescents had depressive symptoms (CES-D score >16), and 5.25% had severe depressive symptoms (CES-D score >28) at baseline. The percentage of the participants had severe depressive symptoms increased to 5.90% in 2021 and 6.94% in 2022. The prevalence of persistent depressive symptoms and severe depressive symptoms were 7.46% and 0.74%. Furthermore, 15.76% adolescents who had depressive symptoms in 2020 aggravated to severe depressive symptoms in 2021 or 2022. Adolescents from urban areas were more likely to have depressive symptoms progression than those from suburbs. The incidence of (severe) depressive symptoms among senior high school students was higher than other students. Girls and adolescents who used screen in the dark, screen time >2 hours/day, and had insufficient sleep were more likely to have depressive symptoms in various outcomes.

In the fully adjusted model, using screen in the dark was significantly associated with higher likelihood of depressive symptoms (OR = 1.89, 95%CI: 1.67, 2.15; P < 0.001) and severe depressive symptoms (OR = 1.89, 95%CI: 1.54, 2.31; P < 0.001) ( Figure 2 ). Compared to screen time ≤ 2 hours/day group, screen time >2 hours/day group was associated with higher likelihood of depressive symptoms (OR = 1.28, 95%CI: 1.15, 1.42; P = 0.002) and severe depressive symptoms (OR = 1.24, 95%CI: 1.03, 1.49; P = 0.025). In addition, insufficient sleep group was associated with higher likelihood of depressive symptoms (OR = 1.56, 95%CI: 1.40, 1.72; P < 0.001) and severe depressive symptoms (OR = 1.84, 95%CI: 1.50, 2.27; P < 0.001) compared with adequate sleep group. Results from unadjusted and adjusted models were generally consistent ( Figure 2 ).

Table 3 illustrates the joint effect of using screen in the dark and screen time > 2 hours/day or insufficient sleep on depressive symptoms. We combined “using screen in the dark + insufficient sleep” and “using screen in the dark + screen time > 2 hours/day” into four groups respectively. Group 1 was the reference group, while Group 4 represented the joint effect. After adjusting for covariates, we found that OR of using screen in the dark and screen time >2 hours/day (OR = 2.22, 95%CI: 1.90, 2.60) on depressive symptoms was slightly higher than only screen time >2 hours/day (OR = 1.24, 95%CI: 1.12, 1.37) or only using screen in the dark (OR = 2.15, 95%CI: 1.73, 2.67). Furthermore, the OR for the combination of using screen in the dark and insufficient sleep (OR = 3.72, 95%CI: 2.79, 4.95) showed a substantial increase in relation to severe depressive symptoms when compared to using screen in the dark alone (OR = 2.09, 95%CI: 1.39, 3.14) or insufficient sleep alone (OR = 2.06, 95%CI: 1.65, 2.57). Likewise, the joint effects of using screen in the dark and screen time > 2 hours/day or insufficient sleep on severe depressive symptoms were consistent with depressive symptoms ( Table 3 ).

In this population-based, three-year longitudinal study conducted among 3,659 adolescents in Shanghai, China, it was found that 23.59% of participants exhibited depressive symptoms, while 5.25% experienced severe depressive symptoms in 2020. 15.76% adolescents who had depressive symptoms in 2020 aggravated to severe depressive symptoms in 2021 or 2022. Female adolescents and senior high school students were more susceptible to depression. Notably, over 10% of the adolescents reported using screens in the dark, a behavior significantly associated with depressive symptoms. Moreover, a robust association was observed between adolescents who used screens in the dark and those who experienced depressive symptoms, particularly when screen time exceeded two hours per day or when there was insufficient sleep, with the association being especially pronounced in cases of severe depressive symptoms.

The prevalence of depressive symptoms among adolescents observed in our study aligns with findings from a global meta-analysis (23), but lower than a national survey conducted at the early stage of the COVID-19 outbreak (24). The reasons might be that the epidemic has a greater impact on adolescents’ mental health, as well as the measurement instruments and cut-off scores were different in these studies. The CES-D, which was utilized in our research, is acknowledged as one of the most extensively used self-assessment tools in psychiatric epidemiology. It has also been widely applied in subsequent studies to screen for depression in adolescents. Previous studies have reported cut-off scores ranging from 14 to 30, exhibiting considerable variability without a consistent pattern (19). This variability complicates comparisons of prevalence rates across studies. Future research should focus on establishing standardized cut-off values to improve cross-study comparability.

Univariate analyses indicated that female students, screen use in dark environments, and inadequate sleep were correlated with depressive symptoms, including severe and persistent forms. High school seniors were at a heightened risk for persistent and cumulative depressive symptoms. This increased risk is attributed to the substantial academic pressure faced by high school seniors while preparing for the National College Entrance Examination, which is considered the most critical examination in their academic careers. In contrast, junior high school and vocational middle school students are exposed to comparatively lower academic demands. Consistent with previous research, our study identified a higher prevalence of depression among female adolescents (27.26%) compared to males (20.18%) in 2020. A national survey indicated that 39.5% (25) of female adolescents aged 11–18 years experienced depression, a figure exceeding that of our study, likely due to its timing during the COVID-19 outbreak in February 2020. The gender disparity in adolescent depression is a well-established phenomenon. Depression arises through multiple pathways, including biological vulnerabilities [e.g., hormonal and neurodevelopmental changes (26)] and psychosocial risk factors [e.g., low self-esteem, exposure to stress and violence (5, 27)]. Nonetheless, depression in males should not be overlooked, as the gender disparity in prevalence diminishes and stabilizes in adulthood (28). Furthermore, additional research is warranted to elucidate the genetic and social determinants underlying these gender differences in depression.

Our study identified an independent association between screen use in dark environments and depressive symptoms in adolescents. As previously mentioned, screen use in the dark may lead to increased LAN exposure. Animal studies (29–31) have demonstrated that LAN exposure can induce anti-social, anxiogenic, and depressive behaviors, even at minimal light levels as low as 5 lux. Human studies further support the LAN-depression link (32–34). A recent study confirmed that LAN is associated with mood and anxiety disorders among adolescents in the United States (4). While, the mechanisms underlying the association between screen use in dark environments and depression remain uncertain. Numerous studies have indicated that exposure to screen light during nighttime may diminish the secretion of melatonin, a hormone integral to sleep regulation, thereby disrupting the circadian timing system (35, 36). Recent research suggests that even brief exposure to light at night (0.5-3 hours) can increase levels of corticotropin-releasing hormone (CRH) and cortisol, potentially leading to heightened anxiety (37). In summary, nighttime light exposure may contribute to emotional disorders through multiple pathways: (1) dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis, (2) suppression of melatonin secretion, and (3) impaired function of mood-related brain regions. Notably, adolescents show greater physiological sensitivity to LAN than adults (38, 39). The delayed melatonin rhythms and stronger suppression of melatonin by nighttime light may explain the higher prevalence of depression and anxiety in adolescents compared to adults.

More than 20% of adolescents reported engaging with electronic devices for several hours daily, a prevalence rate consistent with a study conducted in Guangzhou (6). Many studies (18, 40) have documented the adverse effects of prolonged internet use on adolescent depression. Excessive screen time has been associated with various psychological issues, including low self-esteem, stress, anxiety, depression, insecurity, and loneliness (16, 41). Several researchers have demonstrated that time spent on screens can displace participation in critical activities that may mitigate symptoms of depression, including adequate sleep (6, 42), physical exercise (43), and interpersonal communication with parents and peers. Empirical studies have confirmed the significant mediating role of these factors in the relationship between screen time and depression. Furthermore, excessive use of digital media screens may lead to various adverse effects, such as unhealthy eating behaviors, addiction, poor academic performance, cyber victimization, substance abuse, and diminished social interactions with others (44–46). These adverse effects may also contribute to mental health issues. Conversely, certain studies have indicated that internet use may alleviate depression in adolescents (47), potentially due to participation in recreational activities such as gaming, watching films or short videos, and listening to music, which can aid in stress relief and relaxation (48). Further research Future research should delineate the dual roles of screen media by jointly analyzing usage duration, content type, and contextual factors in adolescent mental health outcomes.

The current study suggests that adolescents who engage in screen use in dark environments coupled with insufficient sleep or prolonged screen time (> 2 hours/day) exhibit a higher likelihood of experiencing depressive symptoms. Notably, the combined effect of screen use in the dark and inadequate sleep significantly related to the severity of depressive symptoms. Furthermore, sleep disorders have been identified as definitive risk factors for depressive symptoms (42, 49). Numerous additional factors may indirectly influence depression by impacting sleep. A mediation analysis study indicated that electronic devices use exerted indirect effects on depressive symptoms through sleep duration reduction among adolescents (50). Nevertheless, the underlying pathophysiological mechanisms linking the combined effects of screen use in dark environments with insufficient sleep or extended screen time to depression remain inadequately understood. Additional high-quality research is required to elucidate these mechanisms.

This study has several limitations. Firstly, residual confounding may persist due to unmeasured variables such as dynamic parenting styles and fluctuating academic stress, which should be modeled as time-varying covariates in longitudinal designs. However, we adjusted for key confounders (e.g., one-child status and academic performance), strengthening the internal validity of our conclusions. Additionally, chronotype misalignment in adolescents may lead to circadian disruption, which could result in the emergence of more severe mood symptoms. It’s needed to investigate chronotype of adolescents and explore the relationship between using screen and depression. Secondly, the data utilized in this study were derived from the Surveillance of Students’ Common Diseases and Health Influencing Factors in Shanghai, China, which was not originally designed as a cohort study. We extracted data spanning from 2020 to 2022. Therefore, the representation of the samples and the resulting findings should be interpreted with caution. Because the economic level of Shanghai is higher, the results may not represent adolescents from cities with lower economic level. What’s more, the causal association and the moderate effect need to explore in the future. Thirdly, the data concerning screen use in low-light conditions were not collected with high precision. It would be advantageous to employ devices to measure screen luminance and to gather detailed information regarding screen types and the content viewed by adolescents in such conditions. The primary challenge was the large-scale nature of the survey, coupled with the fact that the participants were students, which rendered precise measurement difficult. In our future research, we aim to optimize the study design to address these limitations.

In conclusion, this longitudinal study establishes screen use in dark environments—a behaviorally modifiable source of nighttime light exposure—as an independent influencing factor for adolescent depressive symptoms. Adolescents engaging in screen use in the dark, particularly those experiencing insufficient sleep or exceeding two hours of screen time per day, exhibit a higher likelihood of developing depressive symptoms. Concurrently, there is a need for high-quality research to elucidate the causal association and its mechanisms. If these findings hold true, minimizing screen exposure in dark settings may represent a practical strategy for mitigating depression risk. It is imperative for governmental bodies to implement policy measures aimed at reducing nighttime light exposure and enhancing the mental health of adolescents. Greater attention should be directed by educational institutions and society towards addressing depression among female and senior high school students.