Comprehensive effects and dose–response relationship of screen exposure on university students' physical and mental health and academic performance in the digital age: a systematic review and meta-analysis

Abstract

Objective

This study aims to systematically review the impact of total screen time on university students' health outcomes and academic performance, and to explore its dose–response relationship.

Methods

The PubMed, Web of Science, Embase, Cochrane Library, and PsycINFO databases were systematically searched to collect studies related to university students' screen time, with the search period from inception to July 2025. After two researchers independently screened the literature, extracted data, and assessed quality, meta-analysis was performed using Stata 17.0 software. The pooled effect sizes were the reported odds ratios (OR) with 95% confidence intervals (CI). Additionally, dose–response analysis was performed using R 4.4.3 software, and restricted cubic splines were used to fit the non-linear dose–response curve at fixed knots of the exposure distribution.

Results

A total of 69 cross-sectional studies and 3 longitudinal studies were included, comprising 71,633 university students. Meta-analysis results showed that screen time was significantly associated with an increased risk of depression (OR = 1.93; 95%CI: 1.75–2.12) and digital eye strain (OR = 1.96; 95%CI: 1.67–2.31). Dose–response analysis further revealed that after daily screen time exceeded 2.5 h, the risks of depression and digital eye strain increased sharply, with the risks increasing by more than 15% for each additional hour of screen exposure. In addition, screen time was also significantly associated with anxiety, stress, physical pain, sleep quality deterioration, insufficient physical activity, and reduced academic performance, with OR values ranging from 1.12 to 1.79. However, no significant associations were found between screen time and sleep duration or body mass index (BMI).

Conclusions

Current evidence indicates that screen time exerts adverse effects on university students' mental health, physical pain, sleep quality, digital eye strain, physical activity, and academic performance, while no significant association is observed between screen time and sleep duration or BMI.

Protocol registration

The review protocol has been registered at the PROSPERO website since 26 May 2025 (CRD420251060967).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-27025-9.

Introduction

Digital media (smartphones, computers, tablets, gaming consoles, etc.) have been deeply integrated into the daily lives of university students, serving as a core carrier for their knowledge acquisition, social interaction, and leisure activities. Among these devices, smartphones are particularly prevalent. As of 2025, global smartphone users have exceeded 7 billion (covering 85% of the global population), with the ownership rate of smartphones among university students reaching as high as 99.70% [1, 2]. However, this widespread adoption of smartphones is accompanied by severe screen dependency dilemmas. Multiple independent studies from the United States [3], Germany [4], China [5], and India [6] have shown that the average daily screen time of university students is as high as 5.5 to 11 h, far exceeding safe thresholds for screen time recommended by multiple studies.

Excessive use of screen devices exerts multiple negative effects on university students' physical and mental health and academic performance, among which sleep disturbances are particularly prominent. Prolonged screen exposure not only directly impairs sleep quality by inhibiting melatonin secretion and disrupting circadian rhythms, but the cognitive and emotional activation induced by screen content also further delays sleep onset [7, 8]. Sleep disturbances are not only independent health risk factors but also key mediating factors that induce and exacerbate mental health issues such as depression, anxiety, and stress, with their indirect effects accounting for 13.7% to 19.7% of the total effect [9, 10]. Notably, screen activities such as nighttime use, social media browsing and gaming exert a more significant disruptive effect on sleep, with an obvious dose–response relationship observed [11]. In addition, sedentary screen-based behavior also displaces physical activity through the "time substitution effect" [12]. Research indicates that among adolescents with daily screen time exceeding 4 h, the moderate-to-vigorous physical activity compliance rate (≥ 60 min/d) is only 23%, which is significantly lower than that of the group with daily screen time less than 2 h (68%) [13].

Excessive screen device use also directly impairs cognitive functions and hinders academic performance. According to the limited cognitive resource theory, overuse of screen devices tends to cause cognitive overload, attention fragmentation and task interruption, which weakens memory and comprehension efficiency, reduces the depth and continuity of learning, and thus adversely affects academic performance [14]. Meanwhile, prolonged screen use postures such as bowing the head and leaning forward are common triggers for musculoskeletal pain in the neck and shoulders. Related reports indicate that 84% of Canadian university students experience pain due to smartphone use, with the neck being the most commonly affected area [15]. Poor posture, improper ergonomics, and extended periods of screen use not only exacerbate musculoskeletal strain but also easily induce visual fatigue and various ocular symptoms. Surveys indicate that the prevalence of Computer Vision Syndrome (CVS) among university students ranges from 53.3% to 89.9%, with 94% of university students experiencing at least one ocular symptom and 67% reporting three or more symptoms (such as dryness, fatigue, blurred vision, and aching eyes) [16–18]. Thus, It is thus evident that multiple risks, including sleep disturbances, psychological distress, physical inactivity, poor academic performance, musculoskeletal pain, and visual fatigue, are intertwined and jointly compromise university students' learning efficiency, overall health, and subjective well-being.

With the continuous climb in university students' digital media usage, issues such as compromised health and diminished academic performance have become increasingly prominent, evolving into a pressing public health and educational concern. However, the current research landscape exhibits significant limitations, which hinder our comprehensive understanding of these complex issues. First, the research focus is fragmented. Existing studies predominantly concentrate on isolated health indicators (such as sleep quality, mental health, and visual syndrome), making it difficult to reflect the comprehensive impact of screen time on university students' overall health status and academic performance. Second, there is heterogeneity in research findings. Studies on certain health indicators (such as sleep duration, BMI, and physical activity levels) have shown significant inconsistencies, resulting in ambiguous conclusions. Finally, there is insufficient control over key moderating variables. Existing studies often fail to systematically control for potential moderating variables such as study period (COVID-19 pandemic period vs non-pandemic period), screen device type (smartphone vs computer vs various), and population (Asia vs Africa vs South America vs Europe and America), potentially leading to biased effect estimates. Based on the limited cognitive resource theory and the Ecological Systems Theory, this study adopts a dual perspective encompassing cognitive mechanisms and systemic interactions to systematically demonstrate the necessity of integrating multidimensional evidence for clarifying the complex effects of screen use. In response to the current research challenges such as fragmented evidence, inconsistent conclusions, and a lack of unified guidelines for risk assessment and intervention, this study employs meta-analysis to synthesize fragmented research evidence. This approach not only quantifies the overall effect of screen time on university students' health outcomes and academic performance, but also identifies the sources of heterogeneity through subgroup analysis and meta-regression, thereby systematically explaining the inconsistencies in existing research. The findings of this study will provide key evidence-based decision support for developing an integrative theoretical model, formulating educational policies, implementing public health interventions, and promoting individual health management practices.

Methods

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The completed PRISMA checklist is available in Supplement file.

Protocol and registration

The protocol for this systematic review, which aims to evaluate the impact of screen time exposure on university students' health outcomes and academic performance, was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 26 May 2025 (registration number CRD420251060967; available from: https://www.crd.york.ac.uk/PROSPERO/view/CRD420251060967).

Inclusion and exclusion criteria

Studies meeting the following criteria were included: (1) cross-sectional or longitudinal studies (including cohort studies) published in peer-reviewed journals; (2) studies reporting measures of association between screen time and academic performance or health outcomes (including depression, anxiety, stress, physical pain, sleep quality, sleep duration, digital eye strain, BMI, and physical activity), such as odds ratios (OR), relative risks (RR), correlation coefficients (r), or regression coefficients (β); (3) studies published in English; (4) study participants being currently enrolled university students.

The exclusion criteria were: (1) studies that did not report any relevant statistical data; (2) studies focusing on screen content (such as violent video games) or electromagnetic radiation; (3) having severe diseases that may affect the study results (e.g., ocular diseases, mental disorders, and various chronic diseases); (4) chronic use of any medication that could potentially interfere with the study outcomes, including but not limited to psychotropic agents and sedative-hypnotic drugs; (5) reviews, conference abstracts, letters, and duplicate studies.

Search strategy

We systematically searched the PubMed, Web of Science, Embase, Cochrane Library, and PsycINFO databases from inception to July 2025 for relevant studies on screen time among university students. The following MeSH terms and their combinations were used in the search: college student, university student, tertiary student, undergraduate, screen time, digital screen, television, computer, smartphone, video game, and electronic game. Using PubMed as an example, the search strategy is shown below: ((screen time[MeSH Terms]) OR (digital screen[MeSH Terms]) OR (television[MeSH Terms]) OR (computer[MeSH Terms]) OR (smartphone[MeSH Terms]) OR (video game[MeSH Terms]) OR (electronic game[MeSH Terms])) AND ((college student[MeSH Terms]) OR (university student[MeSH Terms]) OR (tertiary student[MeSH Terms]) OR (undergraduate[MeSH Terms])). Additionally, we manually searched the reference lists of included studies, topical reviews, and related articles to identify additional eligible studies. The complete search strategy is available in Supplement file.

Study selection and data extraction

Two researchers independently screened the literature, extracted data, and cross-checked the results. Any discrepancies were resolved through discussions or by consulting a third researcher. After duplicates were removed, obviously irrelevant studies were excluded based on titles and abstracts, followed by a full-text review to determine the final included studies. The extracted data included: title, authors, country/region, publication year, study design, sample size, age, total screen time (hours/day), screen type, screen time categories, confounding factors, and outcome measures.

Certainty of evidence assessment

The certainty of evidence for the included studies was assessed using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) framework, which was rated across four levels: very low, low, moderate, and high. The initial certainty of evidence from cross-sectional and longitudinal studies was rated as "low", and then upgraded or downgraded based on the risk of bias (ROB), indirectness, inconsistency, imprecision, or publication bias. The detailed assessment methods and operational criteria are documented in the supplement file.

Statistical analysis

Meta-analyses were performed using Stata software (version 17.0). The reported OR with its 95%CI was adopted as the pooled effect size. The most appropriate effect sizes adjusted for potential confounders were used if available. For studies that did not report ORs and 95% CIs, we converted the relevant data using multiple methods. For studies reporting regression coefficients, we converted them to ORs with 95%CIs using the method described by Hosmer [19] and Newton [20]. For studies reporting correlation coefficients, we converted them to ORs with 95% CIs using the formula proposed by Borenstein [21]. For studies in which no effect sizes were reported, we manually calculated them using the reported data or available statistics [22, 23]. When studies reported effect sizes separately for different types of screen devices, we combined these effect sizes and used the pooled data for the meta-analysis [24]. For studies that reported effect sizes separately for multiple categories of screen time, we combined effect sizes across these categories using the method proposed by Hamling [25]. If a study provided effect sizes from multiple discrete samples with different population parameters (such as different cohorts), these samples were entered into the meta-analysis separately. Given the heterogeneity in the thresholds used to classify "high" and "low" screen time across the original studies, this meta-analysis used the "lowest exposure group" defined within each study as the unified reference to estimate pooled effect sizes. To account for this and other between-study variations, we used the DerSimonian and Laird random-effects model to estimate an overall OR and 95% CI.

Heterogeneity between studies was evaluated using Cochran's Q test (α = 0.05) and quantified by the I² statistic. Low heterogeneity between studies is indicated when I² ≤ 50% and P ≥ 0.1, whereas high heterogeneity between studies is indicated when I² > 50% and P < 0.1. To test the robustness of the meta-analysis results, sensitivity analyses were performed using the leave-one-out method. Publication bias was assessed by funnel plot asymmetry, and quantified using Begg's test [26] and Egger's test [27]. If significant publication bias was observed, the trim-and-fill method was applied to adjust the pooled results [28]. We further performed subgroup analyses by screen type, study quality, study period, and population to explore potential sources of heterogeneity. Statistical significance was defined as P < 0.05.

This study performed dose–response analyses using R software (version 4.4.3) to investigate the non-linear association between screen time and depression risk. Multiple levels of screen exposure time and their corresponding ORs with 95%CIs were extracted from eligible studies. All screen exposure doses were uniformly converted to "hours/day," with the lowest global exposure dose set as the reference category. To adequately capture potential non-linear relationships between variables, a one-stage restricted cubic spline (RCS) model was adopted and fitted with weighted least squares [29, 30]. Four fixed knots were placed at the 5th, 35th, 65th, and 95th percentiles of the exposure distribution to balance flexibility and stability in model fitting. Owing to insufficient dose–response data for academic performance, anxiety, stress, sleep duration, BMI, and physical activity, dose–response analysis was performed only for depression, physical pain, sleep quality, and digital eye strain.

Results

Study selection

A total of 15,532 records were identified from database searches, and 2 additional records were identified through other sources. After screening against the inclusion and exclusion criteria, 72 studies were included for analysis, including 69 cross-sectional studies and 3 longitudinal studies. These 72 studies corresponded to 73 independent research datasets, because Study [3] included two independent datasets. The study selection process and results are illustrated in Fig. 1.

Fig. 1.

Flowchart of the study selection

Study characteristics

A total of 72 independent studies (73 independent research datasets) from 27 countries were included in this systematic review. These studies were published between 2009 and 2024, comprising 71,633 university students (including 28,845 males, 40,146 females, and 2642 students with unreported gender), all of whom had no prior history of the target outcomes (e.g., mental disorders, ocular diseases, and various chronic diseases) or long-term medication use for these conditions. Among these studies, Regarding mental health, 32 studies used standardized scales to assess the associations between screen time and depression (17 studies) [3, 9, 10, 31–43], anxiety (9 studies) [3, 9, 10, 32, 35, 36, 40, 44], and stress (6 studies) [3, 32, 45–47]. Regarding sleep outcomes, 8 studies assessed sleep duration (via self-report) [4, 6, 31, 34, 48–51], and 25 studies assessed sleep quality (primarily using the Pittsburgh Sleep Quality Index (PSQI)) [3, 4, 6, 7, 9, 10, 34–36, 39, 40, 46–49, 51–59]. In terms of somatic symptoms, 9 studies explored the association between screen time and physical pain via self-report [58–66]. For functional and behavioral indicators, 4 studies evaluated the correlation between screen time and academic performance using grade point average (GPA) [14, 67–69]; 26 studies assessed digital eye strain with standardized tools such as the CVS questionnaire [7, 16, 17, 57, 58, 62, 63, 68, 70–87]; and 5 studies evaluated physical activity level using the International Physical Activity Questionnaire (IPAQ) [3, 12, 88, 89]. In addition, another 5 studies analyzed the correlation between screen time and BMI [3, 90–92]. The studies were conducted in the following regions: Asia (n = 53), Europe (n = 5), South America (n = 3), Africa (n = 5), and North America (n = 7). The basic characteristics of the included studies are presented in Table 1. The GRADE evidence quality assessments and Joanna Briggs Institute (JBI) ROB evaluation results are shown in Supplement file (Tables S1-S3).

Table 1.

Included studies characteristics

Study (Year)	Survey time	Design	Country	Age (year)	Sample size (n)	Screen type	Screen time measurement	Screen time categories (hours/day)	Health outcomes	Adjustment for confounders	Quality score
Liebig et al. (2023) [4]	2020	CS	Germany	24.40 ± 3.90	415 (M 114, FM 292)	smartphone, computer, tablet	self-report	continuous	⑥ ⑦	not indicated	very low
Wright et al.-NCP (2024) [3]	2020	CS	America	23.77 ± 8.63	367	computer, television, smartphone, tablet	self-report	continuous	② ③ ④ ⑦ ⑨ ⑩	not indicated	very low
Wright et al.-DCP (2024) [3]	2020	CS	America	23.77 ± 8.63	598	computer, television, smartphone, tablet	self-report	continuous	② ③ ④ ⑦ ⑨ ⑩	not indicated	very low
Saxena et al. (2021) [6]	2020	CS	India	20.43 ± 1.35	60 (M 30, FM 30)	smartphone, computer, television, video game	self-report	≤ 10 (R), > 10	⑥ ⑦	not indicated	very low
Akçay et al.  (2021) [48]	2019	CS	Turkey	20.57 ± 1.54	752 (M 327, FM 425)	tablet, smartphone, laptop, iPod touch, handheld game player	self-report	continuous	⑥ ⑦	age, sex, family income	low
Sapci et al. (2021) [14]	2019	CS	America	19.07 ± 1.33	99	smartphone	device log	continuous	①	not indicated	very low
Albalawi et al. (2023) [7]	2022	CS	Saudi Arabia	18–45	1593 (M 558, FM 1035)	smartphone, computer	self-report	< 2 (R), 2–4, 4–6, > 6	⑦ ⑧	not indicated	low
Wu et al. (2015) [9]	2013	CS	China	19.26 ± 1.40	4747 (M 1973, FM 2774)	computer, television	self-report	≤ 2 (R), > 2	② ③ ⑦	residential background, gender, perceived family economy, age, BMI, perceived study burden	low
Wang et al. (2023) [10]	2021	CS	China	18–30	1070 (M 461, FM 609)	smartphone, tablet	self-report	continuous	② ③ ⑦	gender, region, education level, type of profession, cost of living, time spent doing exercise	low
Saavedra Vallejos et al. (2025) [12]	not reported	CS	Chile	22.72 ± 4.66	412 (M 171, FM 241)	computer, smartphone	self-report	continuous	⑩	not indicated	very low
Xu et al.  (2018) [16]	2016	CS	China	19.23 ± 1.43	4786 (M 2228, FM 2558)	computer	self-report	≤ 3 (R), > 3	⑧	not indicated	low
Iqbal et al. (2021) [17]	not reported	CS	Egypt	21.17 ± 1.33	4030 (M 1696, FM 2334)	computer, smartphone	self-report	≤ 1 (R), 2, 3, 4, 5, > 6	⑧	not indicated	low
Garcia et al. (2024) [49]	2020–2021	CS	Brazil	24.50 ± 8.60	771 (M 352, FM 419)	television, computer, video game, smartphone	self-report	≤ 8 (R), > 8	⑥ ⑦	diagnosis of COVID-19, time of isolation, anxiety, depression, skin color, gender	low
Sy et al. (2024) [52]	2021–2022	CS	America	19.90 ± 2.30	993	not reported	self-report	continuous	⑦	not indicated	low
Zeyrek et al. (2024) [53]	2022–2023	CS	Turkey	20.97 ± 3.29	443 (M 121, FM 322)	smartphone	device log	< 2 (R), 2–4, 4–8, > 8	⑦	not indicated	very low
Almutairi et al. (2024) [60]	not reported	CS	Saudi Arabia	20.71 ± 1.74	263 (M 122, FM 141)	smartphone, iPad, laptop	not reported	< 2 (R), 2–5, > 5	⑤	not indicated	low
Sharma et al. (2023) [70]	2022	CS	India	21.00 ± 2.20	345 (M 163, FM 182)	smartphone, tablet, laptop, television, desktop	not reported	≤ 4 (R), > 4	⑧	not indicated	very low
Gadain Hassan et al. (2023) [71]	2022	CS	Sudan	not reported	149 (M 87, FM 62)	not reported	self-report	≤ 5 (R), > 5	⑧	not indicated	very low
Fernandez-Villacorta et al. (2021) [72]	not reported	CS	Spain	34.40 ± 6.10	105	smartphone	not reported	1–3 (R), 4–6, 7–10	⑧	not indicated	very low
Lulla et al. (2023) [73]	2021–2022	CS	India	not reported	271	smartphone, computer, television	not reported	≤ 2 (R), 3–4, 4–5, > 5	⑧	not indicated	very low
Sserunkuuma et al. (2023) [31]	2021–2022	CS	Uganda	23.37 ± 3.38	269 (M 157, FM 112)	smartphone	self-report	continuous	② ⑥	not indicated	very low
Fang et al. (2023) [32]	2022	CS	China	not reported	780 (M 302, FM 478)	smartphone	self-report	< 3 (R), 3–7, > 7	② ③ ④	not indicated	very low
Yang et al. (2022) [33]	2014	LC	China	21 (21, 22)	686 (M 686, FM 0)	computer	self-report	≤ 4 (R), > 4	②	not indicated	low
Islam et al. (2021) [34]	not reported	CS	South Korea	not reported	188 (M 95, FM 93)	smartphone	self-report	< 3 (R), 3–5, > 5	② ⑥ ⑦	not indicated	very low
Hammoudi et al. (2021) [35]	2021	CS	Lebanon	21.14 ± 4.08	591 (M 111, FM 480)	smartphone	device log	≤ 7 (R), > 7	② ③ ⑦	not indicated	very low
Feng et al. (2014) [36]	2011	CS	China	18.90 ± 0.90	1106 (M 635, FM 471)	computer	self-report	≤ 2 (R), > 2	② ③ ⑦	sex, age, maternal education, BMI (kg/㎡)	low
Arshad et al. (2021) [54]	2019	LC	Pakistan	21.89 ± 1.73	242 (M 116, FM 126)	smartphone	device log	continuous	⑦	not indicated	low
Goel et al. (2023) [55]	2021	CS	India	17–25	264 (M 126, FM 138)	smartphone	self-report	continuous	⑦	socioeconomic status, literacy level, personality traits	low
Ma et al. (2020) [56]	2015	CS	China	21.74 ± 3.58	5233 (M 1833, FM 3400)	computer	self-report	≤ 2 (R), > 2	⑦	gender, student type, residential background, alcohol usage, age, tobacco usage, losing weight	low
Kalirathinam et al. (2019) [90]	not reported	CS	Malaysia	mean: 20.17	390 (M 154, FM 236)	smartphone, tablet, laptop, video game, desktop	self-report	≤ 2 (R), > 2	⑨	not indicated	very low
Ng et al. (2017) [67]	not reported	CS	Malaysia	not reported	176 (M 33, FM 143)	smartphone	self-report	continuous	①	not indicated	very low
Gao et al. (2021) [44]	not reported	CS	China	19.40 ± 1.50	278 (M 100, FM 178)	not reported	self-report	< 2 (R), 2–3, 3–5, > 5	③	age, rural or urban residency, gender, individual vs team sports	very low
Tanriverdi et al. (2023) [61]	2021	CS	Turkey	20.60 ± 3.00	569 (M 272, FM 297)	computer, tablet, smartphone	self-report	≤ 3 (R), 4–7, 8–10, > 10	⑤	not indicated	low
Ge et al. (2020) [45]	2017	CS	China	19.93 ± 1.07	1137 (M 392, FM 745)	computer, tablet, smartphone	self-report	≤ 6 (R), > 6	④	age, grade, specialization, home location, sleep duration, monthly living expenses	low
Rosenthal et al. (2021) [37]	2019	CS	America	> 18	426 (M 149, FM 277)	smartphone	self-report	< 3 (R), 3–4, 4–6, 7–17	②	race/ethnicity, employment, age group, sexual orientation, first generation college student status, gender, social ladder, social support	low
Zhang et al. (2021) [38]	2020	CS	China	20.51 ± 1.88	11,787 (M 5056, FM 6731)	computer, television, smartphone	self-report	< 2 (R), 2–4, > 4	②	not indicated	low
Montagni et al. (2016) [39]	2013–2015	CS	France	20.80 ± 2.80	4816 (M 1178, FM 3638)	computer, television, smartphone, tablet	self-report	≤ 0.5 (R), 0.5–3, 3–5, > 5	② ⑦	age, gender, study level, paid activity during studying, parental condition, parental moral support and depression	low
Maqbool et al. (2024) [74]	2022	CS	Pakistan	21.30 ± 2.80	415 (M 264, FM 151)	computer, tablet, laptop, smartphone	self-report	< 4 (R), 4–6, 7–10, > 10	⑧	not indicated	very low
Abudawood et al. (2020) [75]	not reported	CS	Saudi Arabia	19–27	587 (M 268, FM 319)	computer	self-report	1–2 (R), 3–4, > 4	⑧	not indicated	very low
Shantakumari et al. (2014) [62]	not reported	CS	UAE	20.40 ± 3.20	471 (M 311, FM 160)	computer	self-report	< 4 (R), 4–6, > 6	⑤ ⑧	not indicated	very low
Lindo-Cano et al. (2022) [76]	2020	CS	Peru	22.41 ± 4.50	709 (M 339, FM 370)	laptop, smartphone	not reported	1–3 (R), 4–6, 7–10, > 10	⑧	not indicated	very low
Venkateshvaran et al. (2023) [77]	2022	CS	India	20.00 ± 1.45	310 (M 107, FM 203)	desktop, tablet, laptop, smartphone	self-report	< 2 (R), 2–3, 3–4, 4–6, > 6	⑧	not indicated	low
Imran et al. (2022) [78]	not reported	CS	Saudi Arabia	not reported	172 (M 124, FM 48)	desktop, laptop, ipad, smartphone	self-report	≤ 2 (R), 3–4, > 4	⑧	not indicated	very low
Talens-Estarelles et al. (2022) [79]	2021	CS	Spain	21.70 ± 3.70	851 (M 317, FM 534)	computer	self-report	continuous	⑧	not indicated	very low
Alqurashi et al. (2024) [80]	2023	CS	Saudi Arabia	22.20 ± 2.00	457 (M 229, FM 228)	not reported	self-report	< 2 (R), 2–4, > 4	⑧	not indicated	low
Rehman et al. (2023) [68]	2022	CS	Pakistan	21.25 ± 1.67	200	not reported	self-report	≤ 0.5 (R), > 0.5	① ⑧	not indicated	very low
Preoteasa et al. (2024) [81]	2022	CS	Romania	22.70 ± 1.66	274 (M 51, FM 223)	not reported	self-report	1–3 (R), 3–5, 5–8, > 8	⑧	not indicated	low
Sitaula et al. (2018) [82]	not reported	CS	Nepal	21.38 ± 1.33	236 (M 180, FM 56)	computer	self-report	≤ 2 (R), > 2	⑧	not indicated	very low
Maroof et al. (2022) [83]	2021	CS	India	20.70 ± 1.90	337 (M 167, FM 170)	smartphone, computer, tablet	self-report	≤ 4 (R), > 4	⑧	not indicated	low
Gebresellassie et al. (2022) [84]	2019	CS	Ethiopia	21.00 ± 5.29	896 (M 454, FM 442)	desktop, tablet, iPad, laptop, smartphone	self-report	1–3 (R), 3–6, 6–8, > 8	⑧	not indicated	very low
Al Tawil et al. (2020) [85]	not reported	CS	Saudi Arabia	not reported	677 (M 0, FM 677)	not reported	self-report	≤ 5 (R), > 5	⑧	not indicated	very low
Indriyani et al. (2022) [86]	2020	CS	Malaysia	18–23	164 (M 104, FM 60)	computer	self-report	< 2 (R), 2–4, > 4	⑧	not indicated	very low
Wang et al. (2021) [87]	2021	CS	China	19.24 ± 0.85	137 (M 60, FM 77)	computer, television, smartphone, tablet	self-report	≤ 4 (R), 5–6, 7–9, 10–12, > 12	⑧	not indicated	low
Peltzer et al. (2019) [50]	2015	CS	ASEAN	20(18.5, 21.5)	3266 (M 1226, FM 2040)	not reported	self-report	≤ 3 (R), 4–6, ≥ 7	⑥	country, age, sex, wealth status, tobacco use, binge drinking, physical activity, depression, self-reported health status, BMI	very low
Xie et al. (2022) [88]	2021	CS	China	not reported	759 (M 389, FM 370)	smartphone	device log	continuous	⑩	not indicated	low
Muhsin et al. (2018) [89]	not reported	CS	Malaysia	19–23	22 (M 0, FM 22)	smartphone, computer, television	self-report	continuous	⑩	not indicated	very low
Benaich et al. (2021) [91]	2019	CS	Morocco	21(20–22)	438 (M 178, FM 260)	computer, television, smartphone, tablet	self-report	≤ 2 (R), > 2	⑨	not indicated	low
Pellerine et al. (2023) [69]	2021–2022	CS	Canada	22.08 ± 3.66	411 (M 76, FM 335)	not reported	self-report	continuous	①	not indicated	very low
Ballard et al. (2009) [92]	not reported	CS	America	mean: 19.50	116 (M 116, FM 0)	computer	self-report	continuous	⑨	not indicated	very low
Laghari et al. (2023) [46]	2020	CS	Pakistan	17–26	130 (M 72, FM 58)	smartphone, computer, television	self-report	continuous	④ ⑦	not indicated	very low
Cheng et al. (2023) [47]	2022	CS	China	not reported	1441 (M 758, FM 683)	smartphone	self-report	< 1 (R), 1–2, 3–5, > 5	④ ⑦	not indicated	low
Guo et al. (2022) [51]	2019	LC	China	22.3 ± 4.20	41 (M 19, FM 22)	smartphone	device log	< 3 (R), 3–4, 4–5, 5–6, 6–7, > 7	⑥ ⑦	sex, monthly household income, alcohol drinking, smartphone operating system, severity of IGD/PSU, worries about social unrest in Hong Kong and being infected with COVID-19, age	very low
Batham et al. (2023) [57]	2022–2023	CS	India	18–30	637 (M 379, FM 258)	smartphone, computer	self-report	< 2 (R), 2–4, 4–6, > 6	⑦ ⑧	not indicated	low
Wahe ed et al. (2019) [58]	2018	CS	Pakistan	17–19	243 (M 101, FM 142)	smartphone	self-report	≤ 2 (R), > 2	⑤ ⑦ ⑧	not indicated	very low
Kim et al. (2024) [63]	not reported	CS	South Korea	21.10 ± 2.50	365 (M 154, FM 211)	computer, laptop, smartphone, e-book	self-report	≤ 5 (R), 6–9, ≥ 10	⑤ ⑧	Obesity (BMI), smoker, use of contact lenses, use of blue light-blocking glasses, window position, daily regular exercise, height of desk, screen brightness	low
Maayah et al. (2023) [64]	2020	CS	Saudi Arabia	21.00 ± 2.95	867 (M 501, FM 366)	smartphone	self-report	1–2 (R), 3–4, 4–5, 6–7, 8–9, 10–12, ≥ 13	⑤	not indicated	low
Alqassim et al. (2024) [65]	2022–2023	CS	Saudi Arabia	21.50 ± 2.10	504 (M 273, FM 231)	smartphone, computer, television	self-report	≤ 1 (R), 2–3, > 4	⑤	age, gender, academic year, BMI	very low
Abou Hashish et al. (2022) [66]	2021	CS	Saudi Arabia	20–23	353 (M 115, FM 238)	computer, tablet, smartphone	self-report	< 0.5 (R), 0.5–2, 2–4, > 4	⑤	not indicated	very low
İkinci Keleş et al. (2021) [59]	2018–2019	CS	Turkey	21.00 ± 2.20	1019 (M 319, FM 700)	smartphone	self-report	1–4 (R), 4–8, 8–12, > 12	⑤ ⑦	not indicated	very low
Liu et al. (2023)[40]	2022	CS	China	21.47 ± 2.60	1453 (M 663, FM 790)	not reported	self-report	≤ 1 (R), 1–2, 2–3, 3–4, > 4	② ③ ⑦	gender, age, physical activity	low
Koly et al. (2021) [41]	2018	CS	Bengal	22.00 ± 2.20	400 (M 195, FM 205)	not reported	self-report	< 2 (R), 2–4, 4–6, > 6	②	not indicated	low
Tashiro et al. (2022) [42]	2021	CS	Japan	21.36 ± 3.01	484 (M 189, FM 295)	not reported	self-report	continuous	②	gender, living status	low
Zhou et al. (2021) [43]	2021	CS	China	not reported	584 (M 349, FM 235)	not reported	self-report	≤ 1.43(R), > 1.43	②	not indicated	low

① = academic performance, ② = depression risk, ③ = anxiety risk, ④ = stress risk, ⑤ = Physical pain, ⑥ = sleep duration, ⑦ = sleep quality, ⑧ = digital eye strain, ⑨ = BMI, ⑩ = physical activity

Abbreviations: CS Cross Sectional Study, LC Longitudinal Cohort Study, R Reference

Screen time and academic performance

Four studies (involving 886 university students) explored the association between categorical exposure to screen time (high vs. low) and academic performance. Meta-analysis showed that university students with longer daily screen time had a higher risk of academic performance decline (OR = 1.79; 95%CI: 1.29–2.48; I² = 31.4%) (Fig. 2). This indicates that excessive screen device use may directly interfere with university students’ ability to complete academic tasks. Due to the limited number of included studies, subgroup analysis for academic performance was not performed.

Fig. 2.

Forest plot of the association between screen time and academic performance

Screen time and depression risk

Seventeen studies (involving 30,352 university students) reported on the association between screen time and depression risk. The pooled results indicated a significant positive correlation between daily screen time and depression risk (OR = 1.51; 95%CI: 1.36–1.68). However, substantial heterogeneity was observed across studies (I² = 94.40%, P < 0.001) (Fig. 3), suggesting that the findings might be influenced by potential confounding factors or methodological differences. After excluding 5 studies that contributed to heterogeneity through sensitivity analysis, the heterogeneity among the remaining studies completely disappeared (I² = 0%), while the pooled effect size increased to OR = 1.93 (95%CI: 1.75–2.12), indicating that the association remained robust (Figures S1 in Supplement file). Subgroup analysis further confirmed the association between screen time and depression risk, with no significant effect modification observed (Table S4 in Supplement file).

Fig. 3.

Forest plot of the association between screen time and depression risk

Six studies (involving 19,070 university students) were included in the dose–response analysis. The results showed a significant non-linear association between screen time and depression risk, with the dose–response curve exhibiting an S-shaped pattern (Fig. 4). When daily screen time was below 1 h, the risk decreased slowly with increasing screen time. However, when screen time exceeded 1 h, the risk rose continuously with increasing screen time. Notably, beyond 2.5 h of daily screen time, the risk of depression increased significantly, with each additional hour of screen exposure associated with a more than 5% increase in depression risk. When daily screen time exceeded 5 h, the rate of increase in the depression risk slowed.

Fig. 4.

Dose–response curve between screen time and depression risk

Screen time with anxiety and stress

Similar positive associations were observed for anxiety risk (9 studies; OR = 1.44) and stress risk (6 studies; OR = 1.58) (Figs. 5 and 6). Although initial meta-analyses demonstrated considerable heterogeneity (I² = 90.60% and 84.80%, respectively), sensitivity analyses excluding selected studies reduced heterogeneity to acceptable levels for each outcome, while the direction and statistical significance of the pooled effect estimates remained stable, further supporting the link between screen time and the aforementioned psychological risks (Figures S2-S3 in Supplement file). This suggests that screen use may evolve from an initial means of emotional regulation (e.g., stress avoidance) into a source of exacerbated emotional distress, and through triggering social comparison, information overload, and sleep deprivation, it may ultimately establish a self-reinforcing cycle in which screen use and anxiety amplify each other. Due to the limited number of included studies, subgroup analysis for anxiety and stress risk was not performed.

Fig. 5.

Forest plot of the association between screen time and anxiety risk

Fig. 6.

Forest plot of the association between screen time and stress risk

Screen time and physical pain

The meta-analysis on physical pain (9 studies, 4654 university students) showed that the risk of physical pain increased with longer daily screen time (OR = 1.60; 95%CI: 1.19–2.15; I² = 84.50%) (Fig. 7). Even after excluding 3 studies that contributed substantial heterogeneity (I² = 41.70%), the association remained significant (OR = 1.70; 95%CI: 1.31–2.20) (Figures S4 in Supplement file). These findings suggest that sedentary behavior and poor posture associated with screen use are important contributors to musculoskeletal pain in university students, which may subsequently interfere with their academic performance. Due to the limited number of included studies, subgroup analysis for physical pain was not performed.

Fig. 7.

Forest plot of the association between screen time and physical pain

Seven studies (involving 3544 university students) were included in the dose–response analysis. The results showed a significant non-linear association between screen time and physical pain, with the dose–response curve exhibiting a U-shaped pattern (Fig. 8). When daily screen time was below 1.6 h, the risk decreased slowly with increasing screen time. However, when screen time exceeded 1.6 h, the risk rose continuously with increasing screen time. Notably, beyond 3 h of daily screen time, the risk of physical pain increased significantly, with each additional hour of screen exposure associated with a more than 5% increase in the risk of physical pain.

Fig. 8.

Dose–response curve between screen time and physical pain

Screen time and sleep duration

The meta-analysis of screen time and sleep duration (8 studies, 5762 university students) showed that longer daily screen time was associated with shorter sleep duration (OR = 1.27; 95%CI: 1.04–1.55), with substantial heterogeneity (I² = 78.80%) (Fig. 9). Sensitivity analysis identified 3 studies as sources of heterogeneity; after excluding these studies, the association was no longer statistically significant (OR = 1.05; 95%CI: 0.93–1.19) (Figure S5 in Supplement file). Given the limited number of included studies, subgroup analysis for sleep duration was not performed.

Fig. 9.

Forest plot of the association between screen time and sleep duration

Screen time and sleep quality

Compared with sleep duration, the association between screen time and sleep quality proved more robust. A meta-analysis of 25 studies (29,213 university students) showed that longer daily screen time was associated with poorer sleep quality (OR = 1.28; 95%CI: 1.18–1.39; I² = 79.30%) (Fig. 10). After excluding 6 studies with considerable heterogeneity in sensitivity analysis, heterogeneity decreased to 47.60%, and the pooled effect size was slightly attenuated but remained significant (OR = 1.12; 95%CI: 1.06–1.18) (Figure S6 in Supplement file). Despite evidence of potential publication bias (Egger's = 0.017), the positive association persisted after trim-and-fill correction (OR = 1.073; 95%CI: 1.008–1.141), indicating that the findings are relatively robust (Figure S7 in Supplement file). Given that sleep quality is fundamental for maintaining cognitive function and overall health, these results suggest that excessive screen use may persistently disrupt key physiological processes essential for restoring energy and consolidating memory in university students.

Fig. 10.

Forest plot of the association between screen time and sleep quality

Seven studies (involving 10,149 university students) were included in the dose–response analysis. The results showed a significant non-linear association between screen time and sleep quality, with the dose–response curve exhibiting an S-shaped pattern (Fig. 11). When daily screen time was below 1.6 h, the risk decreased slowly with increasing screen time. However, when screen time exceeded 1.6 h, the risk rose continuously with increasing screen time. Notably, beyond 3 h of daily screen time, the risk of sleep quality deterioration increased significantly, with each additional hour of screen exposure associated with a more than 10% increase in the risk of sleep quality deterioration. When daily screen time exceeded 5 h, the rate of increase in the risk of sleep quality deterioration slowed.

Fig. 11.

Dose–response curve between screen time and sleep quality

Subgroup analysis stratified by population showed significant associations between screen time and sleep quality in the Asia group (OR = 1.29; 95%CI: 1.18–1.43), and the Europe and America group (OR = 1.37; 95%CI: 1.09–1.72), whereas no significant association was observed in the South America group (OR = 0.87; 95%CI: 0.68–1.11), likely due to only 1 study in this subgroup (insufficient statistical power). All other subgroup analyses showed statistically significant associations, with no significant effect modification observed (Table S5 in Supplement file).

Screen time and digital eye strain

Among all health outcomes analyzed in this study, the association between screen time and digital eye strain was the strongest. A meta-analysis of 26 studies (19,417 university students) yielded a pooled OR of 2.06 (95%CI: 1.70–2.50; I² = 88.80%) (Fig. 12). After excluding 5 studies that were major contributors to heterogeneity (I² = 48.50%), the pooled result remained largely unchanged (OR = 1.96; 95%CI: 1.67–2.31) (Figure S8 in Supplement file). Although potential publication bias was indicated (Egger's = 0.063), the positive association persisted after trim-and-fill correction (OR = 1.79; 95%CI: 1.49–2.14) (Figure S9 in Supplement file). This suggests that prolonged screen viewing can lead to discomfort such as eye dryness and blurred vision, directly impairing learning comfort and efficiency.

Fig. 12.

Forest plot of the association between screen time and digital eye strain

Seventeen studies (11,593 university students) were included in the dose–response analysis. The results showed a significant non-linear association between screen time and digital eye strain, with the dose–response curve exhibiting a U-shaped pattern (Fig. 13). When daily screen time was below 1.7 h, the risk decreased slowly with increasing screen time. However, when screen time exceeded 1.7 h, the risk rose continuously with increasing screen time. Notably, beyond 2.5 h of daily screen time, the risk of digital eye strain increased significantly, with each additional hour of screen exposure associated with a more than 15% increase in the risk of digital eye strain. When daily screen time exceeded 5 h, the rate of increase in the risk of digital eye strain slowed.

Fig. 13.

Dose–response curve between screen time and digital eye strain

Subgroup analysis stratified by screen type showed significant associations between screen time and digital eye strain in the Personal Computer group (OR = 2.39; 95%CI: 1.64–3.48), and the Multiple Devices group (OR = 2.01; 95%CI: 1.58–2.54), whereas no significant association was observed in the Smartphone group (OR = 1.89; 95%CI: 0.50–7.15). For subgroup analysis stratified by population, a significant association was observed between screen time and digital eye strain in the Asia group (OR = 2.36; 95%CI: 1.81–3.07), whereas no significant associations were observed in the Africa group (OR = 1.32; 95%CI: 0.64–2.70), the South America group (OR = 1.95; 95%CI: 0.96–3.96), and the Europe and America group (OR = 2.05; 95%CI: 0.86–4.86). All other subgroup analyses showed statistically significant associations, with no significant effect modification observed (Table S6 in Supplement file).

Screen time with BMI and physical activity

The meta-analysis (5 studies, 1909 university students) revealed no significant association between screen time and BMI (OR = 1.37; 95%CI: 0.99–1.90; I² = 45.90%) (Fig. 14). For physical activity, the initial meta-analysis (5 studies, 2158 university students) did not show a significant association (OR = 1.89; 95%CI: 0.99–3.60) and exhibited very high heterogeneity (I² = 93%) (Fig. 15). However, after removing one key study, heterogeneity decreased substantially and a significant positive association emerged (OR = 1.29; 95%CI: 1.01–1.65; I² = 31.10%), suggesting that the result may be sensitive to individual studies and should be interpreted with caution (Figure S10 in Supplement file). Due to the limited number of included studies, subgroup analysis for BMI and physical activity was not performed.

Fig. 14.

Forest plot of the association between screen time and BMI

Fig. 15.

Forest plot of the association between screen time and physical activity

Risk of bias

Publication bias assessment showed that funnel plots for most health outcomes were generally symmetrical, and no significant bias was detected by Egger's and Begg's tests. Only the funnel plots for sleep quality and digital eye strain were asymmetrical, and Egger's test indicated potential publication bias; however, the results remained consistent after trim-and-fill correction. Overall, the findings of this study demonstrated good robustness (Figure S11-S33 in Supplement file).

Discussion

This meta-analysis integrated 72 studies (n = 71,633) to systematically evaluate the association between screen time and 10 categories of health outcomes among university students. The results indicated that screen time was significantly positively associated with depression, anxiety, stress, physical pain, digital eye strain, reduced academic performance, sleep quality deterioration, and insufficient physical activity, whereas no significant associations were found with sleep duration and BMI. Dose–response analysis revealed that screen time was associated with depression, physical pain, sleep quality, and digital eye strain in a dose-dependent manner. The associations became notably stronger when daily screen time exceeded 2.5 to 3.0 h. Of these outcomes, the associations of screen time with digital eye strain and depression were particularly prominent. Specifically, when daily screen time exceeded 2.5 h, each additional hour of screen exposure was associated with a more than 15% increase in risk. These findings indicate that university students with higher screen time show stronger co-occurrence associations across multiple health outcomes, warranting focused attention in subsequent research and health education efforts.

Screen exposure affects university students' health through multidimensional pathways, with the underlying mechanisms encompassing three interrelated and synergistic dimensions: psychological, physiological, and behavioral. These mechanisms can operate independently but also interact and accumulate synergistically, jointly mediating the link between screen time and a spectrum of adverse health outcomes, including depression, anxiety, stress, sleep disturbances, digital eye strain, physical pain, and reduced academic performance.

Screen use impacts the mental health of university students through multiple pathways, primarily involving the induction of cognitive bias, the fostering of screen dependency, and the impairment of real-world social engagement. First, the upward social comparison induced by social media environments prompts university students to compare themselves with highly idealized virtual profiles or advantaged groups, thereby triggering self-evaluation bias and a sense of relative deprivation, which ultimately catalyzes the generation and intensification of negative emotions such as depression and anxiety [93]. Second, these negative emotions tend to form a bidirectional reinforcing cycle with screen dependency: students experiencing psychological distress are more inclined to escape reality through screen-based activities. Such avoidance behavior further encroaches on time for healthy social interaction and physical activity, thereby exacerbating emotional deterioration and forming a difficult-to-break, self-perpetuating cycle [37]. Meanwhile, prolonged excessive screen use can gradually impair students’ real-world social skills, leading to reductions in empathy and perspective-taking [94, 95]. Of note, screen-based entertainment (such as gaming and social interaction) represents highly rewarding instant-gratification activities that may hijack the brain's psychological reward pathways, displace study time, distract attention, and foster superficial learning and procrastination, thereby interfering with both mental health and academic performance via cognitive and behavioral pathways [96, 97].

At the physiological level, screen use affects health primarily through three pathways: sleep rhythm, visual function, and physical posture. On one hand, blue light emitted by screen devices can disrupt the body's circadian rhythm by suppressing melatonin secretion, thereby disturbing the natural sleep–wake cycle [98–101]. Engaging in stimulating screen activities such as playing video games or browsing information before bedtime can also induce feelings of tension and excitement, which significantly prolongs sleep onset latency [102]. In particular, violent game content can further elevate psychological arousal and exacerbate sleep onset difficulties [10]. Although some studies suggest that electromagnetic fields emitted by screen devices may interfere with slow-wave activity in sleep electroencephalography (EEG), their clinical significance and underlying mechanisms remain controversial and require further investigation [55].

On the other hand, prolonged screen gazing significantly reduces blink rate and increases the incidence of incomplete blinks. This impairs tear film renewal and accelerates tear evaporation, ultimately compromising tear film stability [16, 73]. Meanwhile, the inherent edge-blurring characteristic of pixelated screen images, combined with insufficient resolution and glare reflection, forces the eyes to constantly engage in high-frequency focusing and dynamic accommodation. This leads to ocular muscle fatigue and visual function dysregulation, manifesting as typical symptoms of digital eye strain such as dry eye, blurred vision, and visual fatigue [72, 75]. In addition, exogenous factors such as close screen-viewing distance, contact lens wear, low-humidity environments (< 40%), and high-temperature exposure can synergistically amplify the risk of digital eye strain by increasing accommodative load and accelerating tear evaporation [103, 104]. Existing research confirms that the combined effect of prolonged screen exposure and contact lens wear significantly increases the risk of CVS [105].

Furthermore, maintaining a fixed posture during prolonged screen use (such as computer-based learning and smartphone browsing) increases spinal loading and impairs postural stability, leading to persistent tension and strain in the neck muscles [106, 107]. Maladaptive postures such as forward head posture and frequent neck flexion place the cervical and thoracic spines in a flexed position. Accumulated over time, this increases pressure on the cervical discs and ligaments, induces abnormal mechanical loading in the cervical spine, and directly leads to musculoskeletal pain including neck-shoulder pain and headaches, ultimately creating a vicious cycle in which postural abnormalities and pain exacerbate each other [108–110].

At the behavioral level, time displacement and cognitive overload constitute the core behavioral pathways through which screen exposure impairs health. Specifically, screen-based activities displace time from essential health-promoting behaviors such as studying, sleeping, physical activity, and real-world social interaction, thereby acting as a foundational mechanism for various health impairments induced by screen exposure. Regarding sleep, although screen use displaces nocturnal sleep time via the time displacement effect, university students have relatively flexible daily schedules and can compensate for the loss by waking up later, taking midday naps, or catching up on sleep on weekends. Consequently, deterioration in sleep quality is more pronounced than a significant reduction in sleep duration [111]. Existing research also confirms that university students sleep an average of 1.3 h longer on weekends than on weekdays [112]. For physical activity, regular exercise can preserve mental health through multiple pathways, including enhanced self-efficacy and improved cognitive function. However, excessive screen exposure displaces exercise time, significantly attenuating this protective effect and exacerbating mental health risks [36, 113]. Evidence indicates that university students who effectively limit screen time while maintaining higher physical activity levels can reduce their risk of psychological distress by 40% [114]. Furthermore, cognitive overload serves as a key pathway directly undermining academic performance. Because the brain has limited cognitive resources for information processing, multitasking (e.g., studying while simultaneously engaging in screen-based entertainment) competes intensely for attentional resources. This not only prolongs task completion time but also substantially reduces learning efficiency, ultimately leading to poorer academic performance [96, 97, 115].

The present study did not identify a definitive causal association between screen time and weight gain among university students. Changes in BMI are regulated by a combination of multiple factors, including dietary intake, physical activity expenditure, and genetic metabolism [90]. Although screen use may crowd out exercise time through the time displacement effect, utilizing screen devices for health-related applications such as fitness instruction, exercise monitoring, or nutritional management can facilitate calorie expenditure and promote healthier eating patterns [116]. In addition, age-related metabolic advantages should not be overlooked. University students typically have a higher basal metabolic rate and greater energy compensatory capacity, which may mitigate the potential weight-related risks associated with screen time to a certain extent.

Subgroup analysis shows that Asians face a higher risk of digital eye strain, which is closely associated with Asians’ physiological structure, as well as the educational models and digital media environment in Asian regions. First, Asians' relatively flat eye sockets and prominent cheekbones result in a larger exposed ocular surface area and wider palpebral fissures, thereby accelerating tear evaporation [117]. Furthermore, multiple studies have confirmed that Asians have a higher prevalence of meibomian gland dysfunction (MGD) and poor stability of the tear film lipid layer. These factors make them more susceptible to ocular surface dehydration in dry environments such as classrooms, thereby exacerbating symptoms of visual fatigue [118]. Second, higher education in Asia is highly competitive, and university students’ screen time is significantly above the global average. Universities in China, South Korea, and Japan widely utilize online platforms such as "Rain Classroom" and "K-MOOC" to deliver instruction and course assessments. Students need to use computers or smartphones for extended periods to complete online courses, final assignments, and digital examinations. A survey in India reported a notable prevalence of inadequate ocular rest among university students, with 92% reporting at least one eye-related symptom during digital device use. However, only 11% of medical students adhered to the "20–20-20" eye protection rule, and more than 70% of university students reduced their eye rest due to fear of affecting academic progress [119]. Finally, Asia serves as a core region for global short-form video production and consumption. In China alone, the number of short-form video creators exceeds 150 million; within YouTube's Southeast Asian ecosystem, there are 7,600 channels with over one million subscribers and 77,000 channels boasting more than one hundred thousand subscribers, significantly extending users’ screen time [120]. This high-frequency, fragmented pattern of short-form video consumption creates an additive effect when combined with academic screen time, collectively increasing the risk of digital eye strain among Asian university students. Consequently, alleviating digital eye strain in this group requires not only improving visual environments and enhancing eye protection practices, but also optimizing learning rhythms and regulating behaviors related to digital device usage.

Integrating the present findings and underlying mechanisms, interventions targeting screen-related health risks among university students should adhere to systematic and multi-level principles. In terms of behavioral interventions, it is necessary to establish a safe daily threshold for recreational screen use (e.g., ≤ 2.5 h), implement segmented learning schedules with regular rest breaks (recommended brief rest every 25–30 min), reduce media multitasking behaviors, and ensure adequate time for regular physical activity to counteract the encroachment of the time displacement effect on study, exercise, and sleep. Regarding physiological protection, efforts should be made to promote evidence-based eye care guidelines and the “20–20-20” eye care rule (every 20 min, look at an object 20 feet away for 20 s to promote tear film repair and ocular muscle relaxation) [76], optimize screen postural and visual environmental settings (e.g., positioning the monitor approximately 15° below eye level, maintaining a viewing distance of ≥ 91 cm for computers and ≥ 40 cm for smartphones), and employ strategies such as blue light mitigation and circadian rhythm regulation to improve sleep quality, alleviate visual fatigue, and reduce musculoskeletal strain [70, 72]. For psychological guidance, training in real-world social interaction skills should be enhanced to correct cognitive biases triggered by upward social comparison and break the bidirectional reinforcing cycle of “emotional distress-screen dependence”. For university students presenting with clinically significant sleep disturbances, chronic pain, or emotional disorders, behavioral interventions should be combined with professional assessment and stepped-care support to ultimately achieve comprehensive, multidimensional prevention and control of screen-related health risks throughout the entire process.

Limitations

This study has several limitations. First, the majority of the included studies employed a cross-sectional design, making it difficult to infer a causal relationship from the reported associations. The study results may be confounded by reverse causation or unmeasured confounding factors. Second, due to the inherent nature of observational studies, it is difficult to completely rule out potential biases caused by other factors. Although some studies have adjusted for potential confounding variables, the methods used for such adjustments have varied across different studies. Third, there are significant differences in the methods used to measure screen time across the included studies. Some studies employed objective measures (e.g., Screen Time API), while others relied on self-reported (e.g., questionnaires). Studies based on self-report are more susceptible to recall bias, which may lead to exposure misclassification and thereby reduces the accuracy of effect size estimation. Fourth, given the heterogeneity in the thresholds for classifying "high" and "low" screen time across the included studies, this meta-analysis set the "lowest exposure group" within each study as the universal reference, yet this approach may introduce residual heterogeneity and thereby compromise the accuracy of the pooled effect estimates. Fifth, few studies were included for some outcome indicators, and this limits the feasibility of subgroup analysis. Finally, some studies did not report effect estimates for multiple levels of screen time, making it possible to conduct dose–response analyses for only a subset of outcome measures, thus failing to cover all study outcomes, which may compromise the comprehensiveness of the research. Future research should therefore prioritize prospective cohort designs, integrate standardized objective measurements, and implement stratified confounding control, to establish the causal pathways and dose–response relationships between screen time and health outcomes.

Conclusions

This systematic review found that daily screen time among university students was significantly and positively associated with multiple health indicators. Specifically, longer screen time was associated with elevated risks of mental health problems such as depression, anxiety, and stress. It was also associated with higher risks of decreased sleep quality, digital eye strain, physical pain, poor academic performance, and insufficient physical activity. In contrast, no significant association was observed with sleep duration or BMI. Dose–response analysis further revealed that daily screen time exceeding 2.5 h significantly increased the risk of depression and digital eye strain, whereas daily screen time exceeding 3 h was associated with a marked increase in the risk of physical pain and deterioration in sleep quality. Notably, causal relationships between the variables could not be established in this study, owing to the predominantly cross-sectional design of the included studies, heterogeneity in screen time measurement methods, and insufficient dose–response data for some outcomes. Overall, the findings of this study provide evidence-based support for relevant academic research and public health practice. Future prospective studies using standardized measures are warranted to clarify the dose–response relationship between screen time and health outcomes in university students.

Abbreviations

OR

Odds ratio

RR

Relative risk

CI

Confidence interval

ROB

Risk of bias

BMI

Body mass index

GPA

Grade point average

CVS

Computer vision syndrome

EEG

Electroencephalography

PRISMA

Preferred reporting items for systematic reviews and meta-analyses

GRADE

Grading of recommendations, assessment, development and evaluation

JBI

Joanna Briggs Institute

MGD

Meibomian gland dysfunction

PSQI

Pittsburgh sleep quality index

IPAQ

International physical activity questionnaire

Authors’ contributions

**Zhaolan Zeng:** study conception and design, literature search and screening, data extraction, GRADE evidence assessment, statistical analysis, and manuscript drafting. **Zeyao Shi:** data extraction, GRADE evidence assessment. **Shulin Hou:** literature search, data extraction. **Jing Yan:** critical manuscript revision. **Ru Yang:** literature search and screening, statistical analysis, and manuscript revision. **Xiaowen Li:** GRADE evidence assessment, data collection oversight, final manuscript editing.

Funding

The authors declare that there is no funding for the research.

Data availability

All related data has been presented within the manuscript. The dataset supporting the conclusions of this article is available from the authors on request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Approval of the final version for publication: all coauthors.

Competing interests

The authors declare no competing interests.