Introduction

Emerging adulthood, aged 18 to 25 years, is a critical developmental period marked by identity exploration, social transition, and heightened vulnerability to environmental stressors. During this stage, individuals navigate critical psychosocial tasks including intimate relationship development, social skill acquisition, and family role redefinition. However, such dynamic transformations are often accompanied by heightened stressors, such as academic demands, employment competition, and shifts in living environments, which may exacerbate vulnerability to mental health disorders. Notably, depression has emerged as a major public health issue within the college student population, manifesting through diminished motivation and cognitive fatigue. A meta-analysis of Chinese cohorts reveals alarming prevalence rates, with 28.4 % of college students exhibiting depressive symptoms, substantially surpassing general population estimates. These symptoms not only impair academic functioning and interpersonal interaction but may escalate to severe outcomes including school withdrawal and suicidal behaviors. Given its peak incidence during this transitional phase and ample public health burden, there is an urgent need to investigate its underlying determinants.

In the current digital society, which was further shaped and intensified by the global COVID-19 pandemic, the daily use of social media applications has become an integral part of college students’ lives. The pandemic and associated public health measures have been widely documented to exert profound effects on the mental well-being of college students globally, including increased rates of depression, anxiety, and sleep disturbances. This unique historical context has positioned digital platforms to play an amplified yet complex role in young adults’ lives. Although it provides college students with platforms for self-expression, information acquisition, and the establishment of social connections, excessive use of social media may increase the risk of emotional distress.

Several studies have examined the association between social media use and depression in adolescents; however, their findings remain bidirectional. A longitudinal and adolescent-based study conducted in the United States, using a combination of family-based interviews and adolescents’ self-reported social media use, reported that adolescents who used social media platforms for more than 3 hours per day were more likely to have internalizing symptoms a year postbaseline follow-up. Similarly, in Brazil, a study employing 3161 college students collecting data through self-administered questionnaires demonstrated that 69% reported moderate-to-severe depressive symptoms. Moreover, the study revealed a direct relationship between the duration of social media use and depressive symptoms.

Conversely, studies by Alsunni and Latif about college students in saudi Arabia demonstrated that the frequency of social media use (daytime and night-time) was not associated with anxiety or depression. Aligning with these null findings, Faranda and Roberts’s study of Facebook users aged 18 to 25 years also found no significant correlation between Facebook use and depressive symptoms. These inconsistencies may be attributed to regional, sociodemographic, and methodological differences. In addition, key limitations of existing studies include: (1) insufficient statistical robustness due to small sample sizes, (2) reliance on self-reported social media use through questionnaire scales, and (3) inadequate consideration of diverse social media platform impacts. To our knowledge, only 1 study has used objective measurement of social media use, focusing on 3 commonly used platforms (Facebook, Instagram, and WhatsApp). However, its small sample size (n=15) underscores the pressing need for large-scale, multicenter studies to elucidate the underlying mechanisms.

Prolonged exposure to electronic devices may disrupt circadian rhythms through various biological pathways, such as interfering with the function of the suprachiasmatic nucleus, the hypothalamic-pituitary-adrenal axis, and the dysregulation of melatonin secretion, thereby affecting sleep quality. Sleep problems such as insufficient sleep and difficulty falling asleep may increase an individual’s risk for symptoms such as nonsuicidal self-injury, suicidal behaviour, depression, and anxiety. Therefore, it is reasonable to hypothesize that sleep quality might play a mediating role in the association between social media use and depressive symptoms. To the best of our knowledge, some studies have investigated the mediating role of sleep quality in this context, but reliance on se

Participants and Procedure

A cross-sectional survey was conducted online among undergraduate students at three universities in China between March and april 2023. A convenience sampling method was employed to recruit participants through online platforms, including QQ, miHoYo Community, NetEase LOFTER, Twitter, Today Campus, and PU Pocket Campus. Later, the use durations in the past week for each category can be calculated respectively. GWI’s latest figures indicate that the typical social media user spends 2 hours and 21 minutes using social media each day []. Taking into account multiple factors such as university course studies and social practise, this study defines excessive social media use as exceeding 24 hours per week on social media platforms, serving as the reference group.

Depressive Symptoms

The Self-Rating Depression Scale,developed by Zung [],was used to measure depressive symptoms. This scale consists of 20 items and assesses 10 positive symptoms and 10 negative symptoms over the past nearly 1 week. The scoring range for all items is from 1 (none or a little of the time) to 4 (most or all of the time).The higher the score, the more severe the depressive symptoms are. A standard score higher than 50 indicates depression. To this study, we created a binary outcome variable by combining no and mild depressive symptoms into 1 reference group, contrasted with moderate-to-severe symptoms. This approach was steadfast by our research focus on identifying factors associated with more substantial depressive symptomatology rather than mild mood variations []. In this study, the Cronbach α coefficient of this scale was 0.88.

Sleep Quality

The pittsburgh Sleep Quality Index, developed by Buysse et al [], is a scale used to assess sleep quality in the past month. The overall Pittsburgh Sleep Quality Index score generated by summarizing the total scores of 7 factors ranges from 0 to 21, and a score of 8 or higher is defined as a sleep disorder. Among them, a sleep latency higher than 30 minutes is defined as prolonged sleep latency [], and a sleep time of less than 7 hours at night is defined as insufficient sleep []. In this study, the Cronbach α coefficient of this scale was 0.85.

Health-Related Characteristics

We also collected information on health-related characteristics, including smoking, drinking, height and weight, and physical activity. Smoking was defined as actively smoking one or more cigarettes within the past 30 days. Drinking was defined as having consumed at least 1 type of alcoholic beverage with an alcohol content exceeding 10 grams within the past 30 days. BMI was calculated by dividing weight (kg) by the square of height (m). The physical activity was measured using the International Physical Activity Questionnaire-Short Form, which captures physical activity across multiple domains: occupational activities, transportation, household and gardening tasks, and leisure-time activities (including sports and exercise). According to the standard cut-off levels for calculating metabolic equivalents,physical activity categories were divided into low,moderate,and high levels [].

Statistical Analysis

in this study, SPSS (version 25.0; IBM Corp) and R (version 4.4.2; R Foundation for statistical Computing) software were used for statistical analysis. For categorical data, frequency-based indicators were adopted for statistical description, while for quantitative data, the description was carried out through mean (SD) or median and IQR.Chi-square test or 2-tailed *t* test was used to compare the distribution of different characteristics. For covariates with missing values constituting less than 5%, simple imputation methods were applied based on the distributional characteristics of the covariates. For continuous variables, missing values were imputed by random sampling from a normal distribution with mean and SD derived from the observed data. For categorical variables, missing values were imputed by random sampling from a uniform distribution based on the probability distribution of the observed categories. Had any cova

Demographic Characteristics of the Study Population

This report details the demographic characteristics of the study population, broken down by gender, race, BMI, area of residence, and family type. Data is presented for the total population, as well as for subgroups within the study.

Gender: The majority of the population is male, with 2384 individuals (32.21%) identifying as male, compared to 372 (29.67%) and 121 (33.2%).

Race: The predominant race within the study population is Han, representing 7200 individuals (97.28%) of the total, with 1211 (96.57%) and 347 (95.1%) in the subgroups.

BMI: The distribution of Body Mass Index (BMI) categories reveals the following: 1443 individuals (19.50%) are underweight, 4473 (60.44%) are normal weight, 919 (12.42%) are overweight, and 566 (7.65%) are classified as obese.Subgroup distributions are 266 (21.21%) underweight, 741 (59.09%) normal weight, 140 (11.16%) overweight, and 107 (8.53%) obese; and 77 (21.1%) underweight, 213 (58.4%) normal weight,48 (13.2%) overweight, and 27 (7.4%) obese.

Area of Residence: A significant portion of the population resides in urban areas,with 3474 individuals (46.94%) identifying in this very way, compared to 594 (47.37%) and 176 (48.2%) in the subgroups.

Family Type: The study population primarily consists of individuals from dual-parent family type 1, with 6500 individuals (87.83%) falling into this category.Dual-parent family type 2 represents a smaller proportion at 300 (4.05%), while single-parent families are also represented. Subgroup data shows 1058 (84.37%) and 305 (83.6%) from dual-parent family type 1, 54 (4.31%) and 15 (4.1%) from dual-parent family type 2.

#### Demographic Characteristics of the study Population

The demographic characteristics of the study population are presented in . The sample consisted of 7,367 adolescents, with an average age of 13.72 ± 1.35 years. The distribution of participants across grades was as follows: 5,788 (78.5%) in grade 7, 1,138 (15.4%) in grade 8, and 441 (6.0%) in grade 9. Regarding gender, 3,788 (51.4%) were male and 3,579 (48.6%) were female.

The racial composition of the sample was predominantly White (5,488 [74.4%]), followed by Black or african American (838 [11.4%]), Hispanic or Latino (648 [8.8%]), Asian (249 [3.4%]), and other (144 [2.0%]). The majority of participants resided in urban areas (4,881 [66.3%]), while 2,486 (33.7%) lived in rural areas.

Family structure varied among participants, with 5,381 (73.0%) reporting a dual-parent family (type 1 or type 2^a), 1,488 (20.2%) reporting a single-parent family, and 498 (6.8%) reporting other family arrangements. The average maternal educational attainment was college degree or above (4,298 [58.3%]), and the average paternal educational attainment was also college degree or above (4,389 [59.6%]).

Regarding lifestyle factors, 1,466 (19.8%) participants reported drinking alcohol, while the majority engaged in low levels of physical activity (5,404 [73.0%]). The distribution of physical activity levels was as follows: low (5,404 [73.0%]), moderate (1,363 [18.4%]), and high (634 [8.6%]).#### Associations of Social Media Use With Depressive Symptoms

The relationship between social media use and depressive symptoms is shown in. there was a positive association between the increase in the duration of social media use and the elevation of the odds of depression. After adjusting for gender and grade (model 1 in ), it might very well be observed that as the duration of social media use extends, the likelihood of depression increased accordingly (P value for trend <.001). After further taking into account race,area of residence,siblings,maternal educational attainment,and paternal educational attainment (model 2 in ),the results remained consistent.In the model that continues to incorporate BMI and physical activity (model 3 in ), although the ORs declined slightly (OR_>48h 1.769, 95% CI 1.303‐2.400), the positive correlation between social media use and depression still existed. Regarding the association between the use of instant messaging-based social media and depression, the results were consistent with the overall use situation (OR_>24h 1.728, 95% CI 1.225‐2.437); though, this was not the case for the use of content-based social media and depression (OR_>24h 1.251,

Social Media Use and Mental well-being: dose-Response Analysis

This table presents a dose-response analysis examining the association between different durations of social media use and mental well-being outcomes.Data is presented as hazard ratios (HR) with 95% confidence intervals (CI). The analysis is stratified by type of social media engagement: overall social media use (a), passive social media use (b), and content-based social media use (h).

Overall Social Media Use (a)

The analysis reveals a significant positive trend between overall social media use and the outcome (P < .001). Compared to reference group (≤7 minutes/day), individuals spending 7-16 minutes/day exhibit an HR of 0.954 (0.664-1.369), 16-24 minutes/day show an HR of 1.392 (0.975-1.987), and those using social media for >24 minutes/day have an HR of 1.738 (1.234-2.448). This indicates a progressively increased risk associated with longer durations of overall social media use.

Passive Social Media Use (b)

Similar to overall use, passive social media use demonstrates a significant positive trend (P < .001).Individuals using passive social media for 7-16 minutes/day have an HR of 0.950 (0.662-1.365), 16-24 minutes/day show an HR of 1.391 (0.974-1.987), and those using it for >24 minutes/day have an HR of 1.721 (1.222-2.426). The hazard ratios are comparable to those observed for overall social media use.

Content-Based Social Media Use (h)

Content-based social media use also exhibits a significant positive trend (P < .001). HRs are 0.979 (0.715-1.342) for 7-16 minutes/day, 1.088 (0.798-1.484) for 16-24 minutes/day, and 1.273 (0.949-1.708) for >24 minutes/day. While the trend is consistent, the magnitude of the hazard ratios appears slightly lower compared to overall and passive social media use.

#### Mediating Effect Analysis

shows the results of the mediating effect analysis of multiple sleep quality indicators in the association between social media use and depressive symptoms. Sleep disorders played a partial mediating role in the association between social media use (including instant messaging-based social media and content-based social media) and depressive symptoms, with the mediating effect values being 24.44%, 24.10%, and 25.25% respectively ().

‎

### Discussion

#### Principal Findings

This study characterized the association between social media use and depressive symptoms in emerging adulthood as being dose-dependent, partially mediated by sleep quality, and varying by platform type. Specifically, our analyses demonstrated that there is a significantly positive correlation between the increased duration of social media use among college students and the rising prevalence of depressive symptoms.Even after adjusting for multiple covariates, this association remains stable.The link between the use of instant messaging-based social media and depressive symptoms is consistent with that of overall use of social media, whereas this is not observed for content-based social media. Restricted cubic spline regression analysis revealed a J-shaped relationship between the overall social media use and depressive symptoms. Moreover, sleep disorders play a partial mediating role between social media use and depressive symptoms, with the mediating effect ranging from 24.10% to 25.25%.

In this study, a significant association between social media use and depressive symptoms was found. This result is consistent with previous studies [-].Yan et al [] surveyed 568 college students using a brief scale to assess depressive emotions and mobile social media use, finding a positive correlation between the intensity of mobile social media use and depressive emotions. Additionally, Lin et al [] investigated the social media use and depression among 1787 adults aged between

Social Media, Sleep, and Mental Health in Emerging Adults

The connection between prolonged social media use and poor mental health outcomes is a significant public health concern for emerging adults. Our research demonstrates the potential benefits of promoting mindful social media use and improving sleep hygiene as preventative strategies. These findings contribute to the growing body of scientific evidence informing global conversations, including regulatory discussions in some countries regarding age-specific social media access. Protecting the mental well-being of young people in the digital age requires a extensive approach, combining individual interventions, parental education, and evidence-based policy-making.

Strengths and Limitations

This study took place in late 2022, as universities in China returned to in-person teaching and campus activities following the pandemic.While the lingering effects of COVID-19 may have influenced baseline levels of depressive symptoms and sleep patterns, our primary goal wasn’t to measure absolute prevalence. Rather, we focused on the relationships between social media use and mental health.The dose-response relationships and the role of sleep quality as a mediator are based on established biological and psychological mechanisms, suggesting these findings about digital behavior and mental well-being are robust and relevant to emerging adults today.

our study has several strengths. first, its a large-scale survey of college students. The substantial sample size and strong representativeness give the findings credibility and allow for broader generalization. Second, we precisely measured overall social media use and the time spent on different platforms using data from mobile phone systems. This approach provides more accurate and objective data than customary self-reporting, minimizing potential biases.

However, this study also has limitations. The cross-sectional design prevents us from establishing causality. Future research should consider longitudinal studies or experimental designs to better understand the direction of these relationships.