Abstract
Introduction
Depressive symptoms are common among university students and are associated with poor sleep quality. However, it remains unclear whether the association between sleep quality and depressive symptoms is direct or operates through quality of life (QOL). Screen time may also be associated with sleep quality.
Purpose
This study examined whether QOL mediates the association between sleep quality and depressive symptoms among medical university students, with screen time specified as an upstream behavioral factor of sleep quality.
Patients and methods
A cross-sectional study was conducted among 150 students aged 19–24 years. Sleep quality, depressive symptoms, and QOL were assessed using the Pittsburgh Sleep Quality Index, Center for Epidemiologic Studies Depression Scale, and WHOQOL-BREF (Taiwan version), respectively. Partial least squares structural equation modeling with 5,000 bootstrap resamples was used to estimate the structural paths and indirect associations.
Results
Longer screen time was associated with poorer sleep quality (β = 0.338, p = 0.001). Poorer sleep quality was associated with lower QOL (β = −0.374, p < 0.001), whilst higher QOL was associated with fewer depressive symptoms (β = −0.614, p < 0.001). Although the total association between sleep quality and depressive symptoms was significant (β = 0.327, 95% bootstrap CI [0.120, 0.477], p < 0.001), the direct association was not (β = 0.097, p = 0.212). The specific indirect association through QOL was significant (β = 0.229, 95% bootstrap CI [0.119, 0.327], p < 0.001), consistent with an indirect-only mediation pattern.
Conclusion
Sleep quality was indirectly associated with depressive symptoms through QOL within the specified structural model, while longer screen time was associated with poorer sleep quality. These findings highlight QOL as a central consideration in understanding the association between sleep quality and depressive symptoms, alongside screen-use behaviour in relation to sleep health. Longitudinal studies are needed to clarify temporal relationships among these variables.
Keywords: sleep quality, quality of life, depressive symptoms, screen time, medical university students
Introduction
Depressive symptoms are common among university students and are associated with impaired academic performance, poorer social functioning, compromised physical health, and an increased risk of suicidal behaviours.1–3 An umbrella review of 65 meta-analyses encompassing more than 8.7 million university students reported pooled prevalence estimates of 35.4% for mild depressive symptoms and 13.4% for severe depressive symptoms.4 A survey in Taiwan similarly found that approximately 18.7% of participating university students experienced clinically relevant depressive symptoms.5 Beyond their psychological burden, depressive symptoms can impair functioning and quality of life (QOL), reflecting a broader clinical and public health burden associated with depression.6 The transition to university involves greater autonomy, increased academic demands, changes in living arrangements and social relationships, and disruption of established routines, which may heighten vulnerability to psychological distress.1,3,4 These challenges may be particularly salient among students at medical universities.7,8 Beyond general tertiary educational stressors, undergraduate medical and healthcare students confront substantial academic demands, long study hours, and irregular schedules, which are strongly linked with sleep difficulties and psychological distress.7 Consequently, identifying the modifiable behavioral and psychosocial pathways associated with depressive symptoms within this high-demand cohort is critical for informing targeted institutional mental health initiatives.3
Sleep quality is a multidimensional construct. It encompasses subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, use of sleep medication, and daytime dysfunction.9 Poor and disrupted sleep is common among university students and may be influenced by academic demands, irregular schedules, social activities, and lifestyle behaviours.10 Sleep patterns may also vary across stages and disciplines of study, making faculty affiliation and year of study relevant contextual factors when examining sleep quality among students at a medical university.11 Poorer sleep quality has consistently been associated with greater depressive symptoms.3,12,13 Sleep disruption may affect emotional regulation and psychological functioning, whereas depressive symptoms may adversely affect sleep initiation, continuity, and restorative quality, suggesting a potentially bidirectional relationship.13,14 Neurobiological processes, including neurotransmitter dysregulation and hypothalamic–pituitary–adrenal axis activation, may also contribute to the association between sleep quality and depressive symptoms.3,14 Evidence from randomised controlled trials suggests that interventions targeting sleep may reduce depressive symptoms and other mental health difficulties.15 Nevertheless, previous research has focused predominantly on the direct association between sleep quality and depressive symptoms, while the psychosocial pathways that may help explain this relationship remain poorly understood.16
Among behavioural factors associated with sleep quality, screen time has attracted attention because of the pervasive use of digital devices in university students’ daily lives.16,17 Screen time refers to the cumulative duration of engagement with electronic devices, including televisions, smartphones, desktop or laptop computers, tablets, and in-class projected media.16,17 Greater screen exposure, particularly during the evening or at bedtime, has been associated with poorer sleep.12,16,17 However, findings have not been consistent; for example, one study of healthcare students found no significant association between screen time and sleep quality.18 Differences in the timing, purpose, and type of screen use may contribute to these variations.12,16–18 Epidemiological evidence supports considering screen time as an upstream behavioural correlate of sleep quality.12
Beyond behavioural factors associated with sleep quality, QOL may represent a psychosocial pathway linking sleep quality and depressive symptoms.19,20 The World Health Organization conceptualises QOL as individuals’ perceptions of their position in life in relation to their cultural context, goals, expectations, standards, and concerns.21 It encompasses physical health, psychological well-being, social relationships, and environmental conditions.21,22 Although QOL and depressive symptoms are related, they represent distinct constructs. QOL is a multidimensional assessment of perceived well-being and functioning across physical, psychological, social, and environmental domains.21,22 While the psychological domain of QOL may include affective experiences, it encompasses broader dimensions of psychological well-being and is not synonymous with depressive symptomatology.22 By contrast, depressive symptoms refer specifically to symptoms associated with depression.23,24 This multidimensional construct is relevant to university students, whose well-being is shaped by academic demands, lifestyle transitions, social adaptation, and environmental resources.19,25 Poorer sleep quality has been associated with lower QOL, including dimensions related to fatigue, concentration, emotional functioning, daily activities, and social participation.19,20 In a large international sample, sleep quality was associated with the physical, psychological, social, and environmental domains of QOL, while depressive symptoms were strongly related to poorer overall QOL.19 This aligns with meta-analytic evidence demonstrating consistent, pervasive associations between depressive symptoms and compromised QOL across university student populations.8 These findings suggest that QOL may capture broader dimensions of functioning relevant to the association between sleep quality and depressive symptoms.
Although the association between sleep quality and depressive symptoms is well established, less is known about whether QOL helps explain this association among university students. QOL may provide a broader perspective on how sleep difficulties relate to students’ physical, psychological, social, and environmental well-being. At the same time, screen time may represent an upstream behavioural factor associated with sleep quality. Given the multidimensional nature of student well-being, which encompasses behavioral factors (eg, screen time), physiological needs (eg, sleep quality), and psychological health (eg, depressive symptoms), addressing these interconnected challenges requires a shift from isolated interventions to comprehensive strategies. Therefore, exploring QOL as an integrative pathway could provide empirical evidence for developing multidisciplinary support programs within medical universities. Accordingly, this study aimed to examine the direct and indirect associations among screen time, sleep quality, QOL, and depressive symptoms among students at a medical university, with QOL as a potential explanatory pathway between sleep quality and depressive symptoms. The following hypotheses were proposed: H1, longer screen time is associated with poorer sleep quality; H2, poorer sleep quality is associated with lower QOL; H3, lower QOL is associated with greater depressive symptoms; H4, poorer sleep quality is directly associated with greater depressive symptoms; and H5, poorer sleep quality is indirectly associated with greater depressive symptoms through lower QOL.
Methods
Study Design
This cross-sectional observational study examined the hypothesised associations among screen time, sleep quality, QOL, and depressive symptoms. Faculty affiliation and year of study were included as exogenous covariates predicting sleep quality.
Sample and Setting
Undergraduate students at a medical university in northern Taiwan were recruited by convenience sampling, consistent with a previous study from the same university.26 Eligible students were aged 18–24 years, had been enrolled at the university for at least six months, and provided written informed consent. Students were excluded if they were pregnant; were taking prescribed sleep medication; or reported a physician-diagnosed psychiatric disorder, sleep disorder, chronic disease, or functional impairment.
Measurements
Screen Time
Screen time was assessed using a structured seven-day, paper-based diary. Participants completed the diary immediately before bedtime each day and recorded the duration, in hours, of television, smartphone, desktop or laptop computer use (including electronic gaming), in-class projected media (included to capture classroom-based screen exposure among university students), and tablet use (including iPads or other 3C devices).16 Academic and recreational screen use were not recorded separately. Mean daily screen time was calculated as the total recorded screen time over seven days divided by seven and expressed in hours per day (hours/day) (Supplementary Table S1).
Sleep Quality
Sleep quality during the preceding month was assessed using the Chinese version of the Pittsburgh Sleep Quality Index (PSQI).9,27 The 19 self-report items yield seven component scores: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, sleep-medication use, and daytime dysfunction. Each component is scored from 0 to 3, with the global score ranging from 0 to 21 and higher scores indicating poorer sleep quality. For classification, scores greater than 5 indicated poor sleep quality, whereas scores of 5 or lower indicated good sleep quality.9,27 The Chinese version has reported reliability coefficients of 0.82–0.83 and a test–retest coefficient of 0.85.27
Depressive Symptoms
Depressive symptoms during the preceding week were assessed using the Chinese version of the 20-item Center for Epidemiologic Studies Depression Scale (CES-D).23,28 Items are scored from 0 to 3, with positively worded items reverse-scored. Total scores range from 0 to 60, with higher scores indicating more severe symptoms. A score of 16 or higher was used to identify elevated depressive symptomatology rather than a clinical diagnosis. The CES-D comprises four dimensions: depressed affect, positive affect, somatic symptoms and retarded activity, and interpersonal difficulties.24 The Chinese version has demonstrated satisfactory reliability and validity among university students.24
Quality of Life
QOL during the preceding two weeks was assessed using the 28-item Taiwan version of the World Health Organization Quality of Life–Brief questionnaire (WHOQOL-BREF).21,22 The instrument contains the 26 WHOQOL-BREF items and two Taiwan-specific items. It assesses four domains—physical health, psychological health, social relationships, and environment. Domain scores were calculated by summing the corresponding item scores, with higher scores indicating better QOL. The total QOL score was calculated as the sum of the four domain scores. The WHOQOL-BREF Taiwan version indicated good validity and reliability.22
Procedures
The Institutional Review Board (IRB) of Tri-Service General Hospital approved the study (IRB No. B202005080). Before recruitment, the principal investigator trained the research personnel in eligibility assessment, informed consent, questionnaire administration, and diary completion. To reduce perceived coercion, trained research assistants conducted recruitment and data collection using standardised procedures. After providing written informed consent, participants completed the questionnaires and maintained the screen time diary for seven consecutive days.
Data Analysis
Analyses were conducted using SPSS Statistics 26.0 (IBM Corp., Armonk, NY, USA) and SmartPLS 4 (SmartPLS GmbH, Bönningstedt, Germany).29 Categorical variables are presented as frequencies and percentages, and continuous variables as means and standard deviations. Pearson’s chi-square test was used for sex and faculty affiliation, and the Fisher–Freeman–Halton exact test for year of study. Independent-samples t-tests were used for continuous variables. Pearson’s correlation coefficients were used to examine associations among screen time, sleep quality, QOL, and depressive symptoms. Distributional characteristics were assessed using skewness and kurtosis. p < 0.05 denoted statistical significance.
Partial least squares structural equation modelling (PLS-SEM) was conducted to assess the measurement and structural models.29,30 Bootstrapping with 5,000 resamples was used to assess structural paths and indirect effects. Faculty affiliation was coded as medicine (0) versus non-medicine (1), combining the other four faculties because of small numbers in some categories. Year of study was coded from 1 (first year) to 5 (fifth year); both variables were included as predictors of sleep quality.
Measurement Model
Sleep quality, QOL, and depressive symptoms were specified as first-order reflective constructs, with their respective component or domain scores serving as indicators. Screen time was modelled as a single observed variable. The reflective measurement models were evaluated for internal consistency reliability, convergent validity, and discriminant validity.30,31 Internal consistency reliability was assessed using Cronbach’s alpha and composite reliability (CR). Values ≥0.70 were considered acceptable, with values of 0.60–0.70 acceptable in exploratory research.31 Convergent validity was assessed using outer loadings and average variance extracted (AVE). Outer loadings of 0.70 or above and an AVE of 0.50 or above were preferred. Indicators with outer loadings between 0.40 and 0.70 were considered for removal only when their deletion improved CR or AVE without compromising the content validity and theoretical integrity of the construct.31 Discriminant validity was assessed primarily using the heterotrait–monotrait ratio of correlations (HTMT), with values below 0.85 considered acceptable for conceptually distinct constructs.31 The Fornell–Larcker criterion and indicator cross-loadings were examined. Measurement invariance was assessed using the first two steps of the measurement invariance of composite models (MICOM) procedure. Configural invariance was ensured by applying identical model specifications across groups, and compositional invariance was assessed using 5,000 permutations.32
Structural Model and Indirect Effects
The structural model was assessed using variance inflation factors (VIFs), standardised path coefficients (β), coefficients of determination (R2), effect sizes (f2), and out-of-sample predictive power (Q2_predict).30,31 VIF values below 5 were considered acceptable R2 values of 0.25, 0.50, and 0.75 indicated weak, moderate, and substantial explanatory power, respectively. f2 values of 0.02, 0.15, and 0.35 indicated small, medium, and large effects, respectively. Q2_predict values greater than zero indicated out-of-sample predictive relevance. Bootstrap 95% confidence intervals were reported alongside p values for the indirect and total effects.30,31
Sensitivity Analyses and Alternative Model
Two sensitivity analyses were conducted. First, the model was re-estimated after excluding the psychological health domain of the WHOQOL-BREF from the QOL construct. Second, the model was re-estimated in the overall sample using the global seven-component PSQI score as a single observed indicator of sleep quality, rather than the four-component reflective specification used in the primary model. An alternative model was tested by reversing the proposed ordering of depressive symptoms and QOL, with paths from sleep quality to depressive symptoms and from depressive symptoms to QOL, while retaining a direct path from sleep quality to QOL. Path coefficients and R2 values were compared descriptively between the two models.
Sample-Size Determination
Sample-size adequacy was evaluated using an a priori power analysis with G*Power (version 3.1.9.2) and the inverse square root method for partial least squares structural equation modelling (PLS-SEM).33,34 First, using the linear multiple regression procedure for a fixed model and R2 deviation from zero, with a medium effect size (f2 = 0.15), α = 0.05, 80% statistical power, and three predictors, the minimum required sample size was 77 participants.33 These three predictors corresponded to the maximum number of predictors of an endogenous construct in the specified PLS-SEM model, namely sleep quality. Second, applying the inverse square root method with α = 0.05 and 80% power, a sample size of 150 was estimated to provide sufficient power to detect structural path coefficients with an absolute magnitude of approximately 0.23 or greater.34 This magnitude was considered an appropriate technical benchmark for assessing the statistical sensitivity of the structural model at the available sample size, rather than a substantive or clinically meaningful threshold. Although the final sample of 150 exceeded the minimum required by the a priori power analysis, smaller associations may not have been adequately powered.
Results
Demographic Characteristics
Data were collected from 1 June 2020 to 31 March 2021, when local COVID-19 transmission in Taiwan remained well controlled and the study university maintained in-person instruction. Of the 267 students who agreed to participate, 17 did not meet the eligibility criteria, two withdrew during data collection, and 98 did not complete all required questionnaires. The final sample comprised 150 participants (Supplementary Figure 1). Completers and non-completers differed significantly in sex and year of study (both p < 0.001). No statistically significant difference in faculty affiliation was observed (p = 0.189; Supplementary Table S2).
In the final sample of 150 participants, mean age was 20.87 ± 1.39 years, 55 (36.7%) were male, and 41 (27.3%) had CES-D scores ≥16. Most were from medicine (59.3%), and the largest proportions were in the second (37.3%) and fourth (34.0%) years of study. Overall, 98 (65.3%) had good sleep quality and 52 (34.7%) had poor sleep quality. The mean global PSQI score was 4.93 ± 2.20. Mean age was 20.89 ± 1.46 years in the good sleep quality group and 20.83 ± 1.25 years in the poor sleep quality group; males accounted for 32.7% and 44.2% of the respective groups. Compared with the good sleep quality group, the poor sleep quality group reported longer screen time, higher CES-D scores, and lower total QOL and scores across all four QOL domains (Table 1).
Table 1.
Participant Characteristics Stratified by Sleep Quality (N = 150)
| Construct Indicators | Sleep Quality | Test Statistic | p | ||
|---|---|---|---|---|---|
| Overall | Good (n = 98) | Poor (n = 52) | |||
| M±SD/N (%) | M±SD/N (%) | M±SD/N (%) | |||
| Age, years | 20.87± 1.39 | 20.89 ± 1.46 | 20.83 ± 1.248 | 0.255 | 0.799a |
| Sex, n (%) | 1.961 | 0.161b | |||
| Male | 55 (36.67) | 32 (32.65) | 23 (44.23) | ||
| Female | 95 (63.33) | 66 (67.35) | 29 (55.77) | ||
| DS, n (%) | 2.125 | 0.145b | |||
| CES-D <16 | 109 (72.67) | 75 (76.53) | 34 (65.38) | ||
| CES-D≧16 | 41 (27.33) | 23 (23.47) | 18 (34.62) | ||
| Faculty, n (%) | 1.208 | 0.272b | |||
| Medicine | 89 (59.33) | 55 (56.12) | 34 (65.38) | ||
| Non-Medicine | 61 (40.67) | 43 (43.88) | 18 (34.62) | ||
| Year of study, n (%) | 5.169 | 0.256c | |||
| First-year | 10 (6.67) | 8 (8.16) | 2 (3.85) | ||
| Second-year | 56 (37.33) | 38 (38.78) | 18 (34.62) | ||
| Third-year | 29 (19.33) | 20 (20.41) | 9 (17.31) | ||
| Fourth-year | 51 (34.00) | 28 (28.57) | 23 (44.23) | ||
| Fifth-year | 4 (2.67) | 4 (4.08) | 0 (0.00) | ||
| ST (Hours per day) | 9.46 ± 2.97 | 9.02 ± 2.68 | 10.29 ± 3.34 | –2.518 | 0.013a |
| PSQI global score | 4.93 ± 2.20 | 3.64 ± 1.18 | 7.37 ± 1.52 | –16.610 | < 0.001a |
| DS (CES-D score) | 11.63 ± 7.28 | 10.66 ± 6.69 | 13.46 ± 8.04 | –2.270 | 0.025a |
| QOL domain total scores | 91.93 ± 12.03 | 93.97 ± 12.03 | 88.08 ± 11.17 | 2.926 | 0.004a |
| QOL domain scores | |||||
| Physical health | 25.88 ± 3.68 | 26.64 ± 3.61 | 24.44 ± 3.40 | 3.625 | < 0.001a |
| Psychological health | 19.99 ± 3.75 | 20.44 ± 3.55 | 19.13 ± 3.99 | 2.051 | 0.042a |
| Social relationships | 13.72 ± 2.07 | 14.00 ± 2.11 | 13.19 ± 1.89 | 2.311 | 0.022a |
| Environment | 32.34 ± 4.54 | 32.89 ± 4.55 | 31.31 ± 4.38 | 2.049 | 0.042a |
Notes: p-valuea by Independent-samples t-test; p-valueb by Pearson’s chi-square test; p-valuec by Fisher-Freeman-Halton Exact Test; PSQI scores ≤5 indicate good sleep quality and scores >5 indicate poor sleep quality; QOL domain scores are raw sums; the total QOL score is the sum of the four domains. CES-D ≥16 indicates elevated depressive symptoms (non-diagnostic).
Abbreviations: ST, screen time; SQ, sleep quality; DS, depressive symptoms; QOL, quality of life.
Bivariate Correlations
Screen time was positively correlated with poorer sleep quality (r = 0.346, p < 0.001). Poorer sleep quality was correlated with lower QOL (r = −0.332, p < 0.001) and more depressive symptoms (r = 0.275, p < 0.01), while QOL was inversely correlated with depressive symptoms (r = −0.618, p < 0.001; Table 2).
Table 2.
Pearson Correlations Among the Principal Study Variables
| Constructs | ST | SQ | DS | QOL |
|---|---|---|---|---|
| ST | 1 | 0.346*** | 0.054 | −0.026 |
| SQ | 1 | 0.275** | −0.332*** | |
| DS | 1 | −0.618*** | ||
| QOL | 1 |
Notes: Values are Pearson’s correlation coefficients (r); **p < 0.01; ***p < 0.001.
Abbreviations: ST, screen time; SQ, sleep quality; DS, depressive symptoms; QOL, quality of life.
Measurement Model
Three PSQI components—sleep duration, habitual sleep efficiency, and use of sleep medication—were excluded owing to inadequate outer loadings. The final sleep quality measurement model retained four indicators with outer loadings ranging from 0.585 to 0.779. Although Cronbach’s α (0.582) and AVE (0.443) fell below prespecified thresholds, composite reliability was acceptable (CR = 0.758). These four sleep quality indicators were retained because their removal did not sufficiently improve the measurement criteria. Outer loadings ranged from 0.741 to 0.883 for depressive symptoms and from 0.815 to 0.877 for QOL. Consequently, the final measurement model comprised four indicators for each latent construct. Depressive symptoms (α = 0.809, CR = 0.874, and AVE = 0.636) and QOL (α = 0.875, CR = 0.912, and AVE = 0.722) met the prespecified criteria for internal consistency and convergent validity (Supplementary Table S3). Discriminant validity was supported, with HTMT values ranging from 0.457 to 0.706; the Fornell–Larcker criterion and indicator cross-loadings provided consistent results (Supplementary Table S4, S5, and S6). Indicator-level VIFs ranged from 1.106 to 2.839, indicating no critical collinearity. Skewness (−0.599 to 1.321) and kurtosis (−0.448 to 2.434) were within the prespecified ranges (Supplementary Table S7).
Structural Model and Indirect Effect
In the overall sample, longer screen time was associated with poorer sleep quality (β = 0.338, p = 0.001). Year of study (β = 0.166, p = 0.083) and faculty affiliation (β = 0.052, p = 0.784) were not significantly associated with sleep quality. Poorer sleep quality was associated with lower QOL (β = −0.374, p < 0.001), and higher QOL was associated with fewer (β = −0.614, p < 0.001). The direct association between sleep quality and depressive symptoms was not significant (β = 0.097, p = 0.212). In contrast, the specific indirect association between sleep quality and depressive symptoms through QOL was significant (β = 0.229, 95% bootstrap CI [0.119, 0.327], p < 0.001), as was the total association (β = 0.327, 95% bootstrap CI [0.120, 0.477], p < 0.001) (Figure 1, and Table 3). These findings indicated an indirect-only mediation pattern, in which the association between sleep quality and depressive symptoms was statistically significant through QOL but not directly.
Figure 1.

PLS-SEM results for the overall sample.
Notes: Faculty affiliation was coded as medicine = 0 and non-medicine = 1; year of study was coded from 1 to 5; Values adjacent to indicators are outer loadings. Values on the structural paths are path coefficients, with p values in parentheses; values within the endogenous constructs are coefficients of determination (R2).
Table 3.
Direct, Specific Indirect, and Total Effects Estimated Using PLS-SEM
| Structural Relationship | β | SE | t | P | 95% CI | f2 | |
|---|---|---|---|---|---|---|---|
| Lower | Upper | ||||||
| Overall | |||||||
| Direct Effect | |||||||
| ST → SQ | 0.338 | 0.098 | 3.438 | 0.001 | 0.132 | 0.513 | 0.128 |
| SQ → QOL | −0.374 | 0.079 | 4.733 | < 0.001 | −0.499 | −0.177 | 0.163 |
| QOL → DS | −0.614 | 0.061 | 10.084 | < 0.001 | −0.718 | −0.477 | 0.552 |
| SQ → DS | 0.097 | 0.078 | 1.247 | 0.212 | −0.057 | 0.249 | 0.016 |
| Specific indirect effect | |||||||
| SQ → QOL → DS | 0.229 | 0.053 | 4.334 | < 0.001 | 0.119 | 0.327 | |
| Total Effect | |||||||
| SQ → DS | 0.327 | 0.088 | 3.710 | < 0.001 | 0.120 | 0.477 | |
Note: Confidence intervals are 95% bias-corrected bootstrap intervals. P values are based on the bootstrap t statistics reported by SmartPLS.
Abbreviations:β, standardised path coefficient; CI, confidence interval; f2, effect size; SE, standard error; ST, screen time; SQ, sleep quality; DS, depressive symptoms; QOL, quality of life.
The model explained 43.1% of the variance in depressive symptoms (R2 = 0.431; adjusted R2 = 0.423) and 14.0% of the variance in QOL (R2 = 0.140; adjusted R2 = 0.134). Effect sizes were large for the QOL to depressive symptoms path (f2= 0.552), medium for the sleep quality to QOL path (f2 = 0.163), and small for the direct sleep quality and depressive symptoms association (f2= 0.016). However, out-of-sample predictive performance was limited. The Q2_predict values were close to zero for depressive symptoms (0.010) and negative for QOL (−0.009), indicating no meaningful out-of-sample predictive relevance for either construct. These findings suggest that the model is better interpreted in terms of its explanatory power than its predictive performance (Table 3 and Supplementary Table S8).
Exploratory Subgroup Analyses
Exploratory subgroup analyses are presented in Supplementary Figure 2 and Supplementary Tables S8–S9. The first two MICOM steps supported configural and compositional invariance for sleep quality, QOL, and depressive symptoms across the good and poor sleep quality groups (Supplementary Table S10). Given the restricted within-group variability, limited subgroup sample sizes, and absence of formal comparisons of structural paths, these results do not establish differences in structural associations between the sleep quality groups.
Sensitivity Analyses and Alternative Model
In the sensitivity analysis, after excluding the psychological health domain of the WHOQOL-BREF from the QOL construct, the association between QOL and depressive symptoms remained significant (β = −0.459, p < 0.001). The specific indirect association between sleep quality and depressive symptoms through QOL was also significant (β = 0.174, 95% bootstrap CI [0.082, 0.276], p = 0.001) (Supplementary Figure 3). When the model was re-estimated using the global seven-component PSQI score as a single observed indicator of sleep quality, the direct association between sleep quality and depressive symptoms remained non-significant (β = 0.061, p = 0.387), whereas the specific indirect association through QOL remained significant (β = 0.209, 95% bootstrap CI [0.114, 0.309], p < 0.001), as was the total association (β = 0.270, 95% bootstrap CI [0.121, 0.412], p < 0.001) (Supplementary Figure 4). In the alternative model, sleep quality was associated with depressive symptoms (β = 0.331, p < 0.001), depressive symptoms with QOL (β = −0.591, p < 0.001), and sleep quality directly with QOL (β = −0.177, p = 0.010). The model explained 11.0% of the variance in depressive symptoms and 45.0% in QOL, compared with 43.1% and 14.0%, respectively, in the proposed model (Supplementary Figure 5). These differences in explained variance do not establish the temporal ordering of the constructs.
Discussion
Three main findings emerged from the overall model. First, poorer sleep quality was associated with lower QOL, and lower QOL was strongly associated with more depressive symptoms. The indirect association between sleep quality and depressive symptoms through QOL was significant, whereas the direct association was not significant. According to the proposed classification,35 this pattern is consistent with indirect-only mediation. Second, the indirect association was significant in the good sleep quality group but not in the poor sleep quality group; however, no formal between-group comparison of structural paths was conducted. Third, longer screen time was associated with poorer sleep quality in the overall model, consistent with the hypothesised relationship between screen time and sleep quality.
Quality of Life as a Central Integrative Pathway
The contribution of the present study lies in testing the simultaneous relationships among screen time, sleep quality, QOL, and depressive symptoms within an integrated structural model, with particular attention to the indirect association between sleep quality and depressive symptoms through QOL. Within this framework, screen time was positioned as an upstream behavioural correlate of sleep quality, while QOL was examined as a potential explanatory pathway linking sleep quality with depressive symptoms.19,20 QOL captures physical health, psychological well-being, social relationships, and environmental conditions,19,20 providing a broader perspective on everyday functioning than sleep quality or depressive symptoms considered separately. Previous research has demonstrated associations between sleep quality, QOL, and mental health among university students, including studies that have examined these factors concurrently within structural models.19,36 Other research has investigated sequential associations involving depression, QOL, and sleep regularity.37 Specifically, Moussa-Chamari et al19 examined the four QOL domains separately and reported indirect pathways linking sleep quality to physical health, social relationships, and environmental QOL. The present study complements this work by examining overall QOL, encompassing all four domains, as a potential explanatory pathway between sleep quality and depressive symptoms, while incorporating diary-based screen time as a behavioural correlate of sleep quality. Poorer sleep quality was associated with lower QOL, consistent with previous evidence linking sleep difficulties to impaired functioning and well-being among university students.36 An alternative model placing depressive symptoms between sleep quality and QOL also yielded statistically significant structural paths. The cross-sectional design precludes determination of temporal ordering or causal mediation in either model. Similar patterns have been reported among adolescents and emerging adults, in whom sleep and other health behaviours are associated with psychological outcomes in relation to everyday functioning and well-being.37
Stress appraisal theory provides a useful framework for interpreting this pattern.38 The theory proposes that psychological responses are shaped by how individuals evaluate their circumstances and available coping resources. Applied to the present findings, poorer sleep quality may be associated with a less favourable appraisal of physical and psychological functioning, social relationships, and the living environment, which may be reflected in lower QOL. Lower QOL was, in turn, strongly associated with greater depressive symptoms. This perspective provides a theoretical interpretation of the observed indirect association between sleep quality and depressive symptoms through QOL, while the direct association between sleep quality and depressive symptoms was not significant. Importantly, appraisal processes were not assessed in the present study; therefore, this framework represents an interpretation of the observed associations rather than a tested mechanism.
The multidimensional nature of QOL may be particularly relevant to this pattern.19,37 Sleep difficulties may be reflected across several aspects of perceived functioning and well-being rather than being expressed only as depressive symptoms.19,25 Thus, examining QOL alongside sleep quality may provide a broader framework for understanding the association between sleep difficulties and depressive symptoms among university students.
Subgroup-Specific Findings
In the exploratory subgroup analyses, the specific indirect association between sleep quality and depressive symptoms through QOL was statistically significant in the good sleep quality group but not in the poor sleep quality group. In the poor sleep quality group, sleep quality was not significantly associated with QOL. Subjective well-being homeostasis theory proposes that well-being may remain relatively stable despite adverse circumstances until an individual’s adaptive capacity is exceeded.39,40 This theory may offer one perspective on the subgroup findings. Previous studies have reported associations between sleep quality and subjective well-being, including among university students.41,42 Among students with poor sleep quality, sleep problems may coexist with academic, psychological, and other daily pressures, potentially making the association between sleep quality and QOL less apparent.
The association between screen time and sleep quality was statistically significant only in the poor sleep quality group. These findings require cautious interpretation because stratifying a continuous measure reduces within-group variability and may weaken statistical power.43 The poor sleep quality subgroup included only 52 participants, potentially limiting the precision and stability of its estimates. Furthermore, statistical significance in one subgroup but not another does not establish a difference between groups; formal between-group comparisons are required.43,44 As no formal comparison of structural paths was conducted, these findings should be interpreted as exploratory and hypothesis-generating. Larger studies with adequate statistical power and formal between-group comparisons are needed to investigate whether these associations vary by sleep quality status.
Screen Time and the Conceptualisation of Sleep Quality
In the overall structural model, longer screen time was associated with poorer sleep quality, consistent with previous studies among university students.16,45,46 Screen use near bedtime may be associated with sleep difficulties through delayed bedtime and increased cognitive or emotional arousal.45,46 Experimental evidence has also suggested that reducing smartphone screen time may improve sleep quality and psychological well-being in students.47 Although causal relationships cannot be determined from the present cross-sectional data, the findings indicate that screen-use habits should be considered when addressing sleep problems among university students.
The PSQI assesses multiple dimensions of self-reported sleep.9 In the present study, the four retained components mainly reflected perceived sleep quality, sleep latency, sleep disturbances, and daytime dysfunction, consistent with findings among medical students.48 These components represent key aspects of subjective sleep experience and its daytime consequences. The exclusion of sleep duration, habitual sleep efficiency, and sleep medication use indicates their lower contribution to the sleep quality construct in this sample rather than a lack of importance. Evidence from psychiatric outpatient settings further highlights the clinical relevance of structured sleep assessment. In an Italian community mental health sample, 40.5% of patients met the study threshold for clinically significant insomnia on the Insomnia Severity Index, and depressive disorders were associated with higher odds of clinically significant insomnia.49
Implications
The observed associations among sleep quality, QOL, and depressive symptoms suggest that these domains may be considered jointly when assessing student well-being. The physical, psychological, social, and environmental domains of QOL provide a broader perspective on students’ everyday functioning.8,19 These findings may inform the development of multidisciplinary assessment and support approaches; the effectiveness of such approaches requires evaluation in intervention studies.
Screen time warrants consideration given its observed association with sleep quality. The present study assessed total daily screen time and did not establish whether changes in screen use would improve sleep or reduce depressive symptoms. Future studies could examine the timing and purpose of screen use and evaluate whether modifying screen-use behaviors improves sleep outcomes.17,46
Limitations and Future Research Directions
This study has several limitations. First, the cross-sectional design precludes conclusions about causality or temporal ordering among screen time, sleep quality, QOL, and depressive symptoms. Variations in the recall periods of the measures (seven days for the screen time diary and CES-D, two weeks for the WHOQOL-BREF, and one month for the PSQI further constrain temporal interpretation. Longitudinal studies are needed to examine temporal relationships, whilst intervention studies could provide evidence regarding causality. Second, the measurement of sleep quality warrants cautious interpretation. Three PSQI components—sleep duration, habitual sleep efficiency, and use of sleep medication—were excluded from the reflective measurement model because of inadequate outer loadings. The retained four-component specification may therefore have limited the content coverage of sleep quality. In the sensitivity analysis using the global seven-component PSQI score, the direct association between sleep quality and depressive symptoms remained non-significant, while the specific indirect association through QOL remained significant (Supplementary Figure 4). This consistency suggests that the observed pattern of associations is not solely dependent on the four-component specification. It does not resolve the limitations in content coverage, internal consistency (α = 0.582), and convergent validity (AVE = 0.443) of the primary reflective measurement model. Potential content overlap between the psychological health domain of the WHOQOL-BREF and depressive symptoms warrants cautious interpretation. After this domain was excluded, the associations between QOL and depressive symptoms, and the indirect association between sleep quality and depressive symptoms through QOL, remained statistically significant, with smaller absolute coefficients (Supplementary Figure 3). These findings suggest that the indirect association is not solely dependent on the psychological domain, although concerns about content overlap remain. Measurement error may also have attenuated associations involving sleep quality. Screen time, sleep quality, QOL, and depressive symptoms were assessed using self-reported measures. The exclusive use of this measurement approach may have introduced common method variance and biased the estimated associations. The direction and magnitude of this potential bias remain uncertain. Future studies could examine alternative measurement specifications and combine self-reported sleep measures with objective assessments, such as actigraphy.50,51 Third, convenience sampling from a single medical university in Taiwan may limit generalisability to students with similar demographic, educational, and cultural characteristics. Excluding students who reported a physician-diagnosed psychiatric disorder may have selected a relatively healthier sample with respect to mental health. This may limit generalisability to the broader student population, particularly students with psychiatric disorders. Multi-centre studies with more diverse student populations are warranted. Data were collected during the COVID-19 pandemic, when the study university maintained in-person instruction. Potential pandemic-related influences on students’ daily routines, screen use, and well-being remain uncertain. Approximately five years and five months elapsed between the end of data collection and initial submission. The findings should be interpreted within the 2020–2021 context, and their applicability to current student populations requires further investigation. Fourth, QOL may not fully account for the association between sleep quality and depressive symptoms. Perceived stress, emotion regulation, resilience, social support, and coping competence were not assessed and may represent additional factors.8,52 The primary model included faculty affiliation and year of study as predictors of sleep quality only. Sex was not included, and faculty affiliation and year of study were not specified as direct predictors of quality of life or depressive symptoms. Potential confounding by these demographic and educational factors should therefore be considered when interpreting the structural associations. Fifth, our definition of screen time included in-class projected media and did not distinguish between academic and recreational use. The diary did not separately record overlapping use of multiple devices, so its contribution to the total recorded screen time could not be determined. This broad definition may limit comparability with studies focusing primarily on personal or recreational screen use. Future studies using standardised measures that account for device type and usage context may help clarify these differences. Sixth, a further limitation is the potential for selection bias due to differential questionnaire completion. Of the 250 eligible students, 100 (40.0%) were non-completers, including 98 who did not complete all required questionnaires and two who withdrew during the study. Completers and non-completers differed significantly in gender and year of study, but not in faculty affiliation (Supplementary Table S2). These differences suggest that the final analytic sample may not fully represent all eligible students. As the study variables were unavailable for non-completers, we could not assess whether questionnaire completion was associated with screen time, sleep quality, QOL, or depressive symptoms. Consequently, the direction and magnitude of potential selection bias remain uncertain, and the findings should be interpreted in the context of the final analytic sample. Finally, the exploratory subgroup analyses require particular caution because stratification by the PSQI global score restricts the variability of sleep-related measures within each group. The poor sleep quality subgroup comprised only 52 participants, which may have limited the precision and stability of the subgroup estimates. In addition, no formal comparison of structural paths between groups was conducted. The subgroup analyses are therefore presented in the Supplementary Information for transparency and should not be interpreted as evidence of differences in the structural associations by sleep quality status. Future studies with larger samples and longitudinal designs could examine whether these associations vary across levels of sleep quality.
Conclusions
An indirect association between sleep quality and depressive symptoms, mediated by QOL, was observed in the specified cross-sectional model. Longer screen time was associated with poorer sleep quality. These findings highlight the relevance of QOL to student well-being and the importance of considering screen-use behaviour in relation to sleep health. Longitudinal studies are needed to clarify temporal relationships, while intervention studies are needed to evaluate the effectiveness of proposed support approaches.
Acknowledgments
The authors extend their deepest gratitude to all the young adult participants who participated in this study. We also wish to express our sincere thanks to the faculty and students at the National Defense Medical University for their invaluable support.
Funding Statement
This research was funded by the Ministry of Science and Technology (Taiwan) [grant numbers MOST 109-2813-C-016-014-B].
Abbreviations
AVE, Average variance extracted; CES-D, Center for Epidemiologic Studies Depression Scale; CR, Composite reliability; HTMT, Heterotrait–monotrait ratio of correlations; IRB, Institutional Review Board; MICOM, Measurement invariance of composite models; PLS-SEM, Partial least squares structural equation modelling; PSQI, Pittsburgh Sleep Quality Index; QOL, Quality of life; VIF, Variance inflation factor; WHOQOL-BREF, World Health Organization Quality of Life–Brief questionnaire.
Data Sharing Statement
The datasets generated and/or analysed during the current study are not publicly available in order to protect the privacy of the study participants, but are available from the corresponding author on reasonable request.
Ethical Approval
The study was approved by the Institutional Review Board (IRB) of Tri-Services General Hospital (IRB No. B202005080). Informed consent was obtained from all subjects involved in the study. All experiments were conducted following the Declaration of Helsinki and relevant national and institutional guidelines and regulations.
Disclosure
The authors declare no competing interests.
References
- 1.Li L, Lok GKI, Mei SL, Cui XL, An FR, Li L. Prevalence of depression and its relationship with quality of life among university students in Macau, Hong Kong and mainland China. Sci Rep. 2020;10(1):15798. doi: 10.1038/s41598-020-72458-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Rosenthal SR, Zhou J, Booth ST. Association between mobile phone screen time and depressive symptoms among college students: a threshold effect. Hum Behav Emerg Technol. 2021;3(3):432–14. doi: 10.1002/hbe2.256 [DOI] [Google Scholar]
- 3.Zhang Z, Cai Z, Meng Q. Negative life events, sleep quality and depression in university students. Sci Rep. 2025;15(1):21193. doi: 10.1038/s41598-025-08635-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Paiva U, Cortese S, Flor M, Moncada-Parra A, Lecumberri A, Eudave L. Prevalence of mental disorder symptoms among university students: an umbrella review. Neurosci Biobehav Rev. 2025;175:106244. doi: 10.1016/j.neubiorev.2025.106244 [DOI] [PubMed] [Google Scholar]
- 5.Ye YX. Barriers to help-seeking for depression across the lifespan. Newsletter Taiwan Assoc Against Depress. 2024;23(1):12–17. [Google Scholar]
- 6.Martiadis V, Demyttenaere K, Martinotti G, Fiorillo A. Editorial: treatment resistant depression: epidemiology, clinic, burden and treatment, volume II. Front. Psychiatry. 2026;17:1935707. doi: 10.3389/fpsyt.2026.1935707 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wondie T, Molla A, Mulat H, et al. Magnitude and correlates of sleep quality among undergraduate medical students in Ethiopia: cross-sectional study. Sleep Sci Pract. 2021;5(1):7. doi: 10.1186/s41606-021-00058-2 [DOI] [Google Scholar]
- 8.Fernandes M, Mendonça CR, da Silva TMV, Noll P, de Abreu LC, Noll M. Relationship between depression and quality of life among students: a systematic review and meta-analysis. Sci Rep. 2023;13(1):6715. doi: 10.1038/s41598-023-33584-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh sleep quality index: a new instrument for psychiatric practice and research. Psychiatry Res. 1989;28(2):193–213. doi: 10.1016/0165-1781(89)90047-4 [DOI] [PubMed] [Google Scholar]
- 10.Schmickler JM, Blaschke S, Mess F, et al. Sleep in the academic sphere: identifying sleep profiles and their influencing factors using latent profile analysis in German university students. BMC Psychol. 2025;13(1):907. doi: 10.1186/s40359-025-03280-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Schmickler JM, Blaschke S, Robbins R, Mess F. Determinants of sleep quality: a cross-sectional study in university students. Int J Environ Res Public Health. 2023;20(3):2019. doi: 10.3390/ijerph20032019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bhandari PM, Neupane D, Rijal S, Thapa K, Mishra SR, Poudyal AK. Sleep quality, internet addiction and depressive symptoms among undergraduate students in Nepal. BMC Psychiatry. 2017;17(1):106. doi: 10.1186/s12888-017-1275-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Angelillo S, Sansone V, Paduano G, Lateano L, Di Giuseppe G, Nobile CGA. The quality of sleep: evaluation among university students. Front Public Health. 2023;11:1270426. doi: 10.3389/fpubh.2023.1270426 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Nicolaides NC, Charmandari E, Chrousos GP, Kino T. Circadian endocrine rhythms: the hypothalamic-pituitary-adrenal axis and its actions. Ann N Y Acad Sci. 2014;1318:71–80. doi: 10.1111/nyas.12464 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Scott AJ, Webb TL, Martyn-St James M, Rowse G, Weich S. Improving sleep quality leads to better mental health: a meta-analysis of randomised controlled trials. Sleep Med Rev. 2021;60:101556. doi: 10.1016/j.smrv.2021.101556 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Mohd Saat NZ, Hanawi SA, Hanafiah H, Ahmad M, Farah NMF, Abdul Rahman NAA. Relationship of screen time with anxiety, depression, and sleep quality among adolescents: a cross-sectional study. Front Public Health. 2024;12:1459952. doi: 10.3389/fpubh.2024.1459952 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kaewpradit K, Ngamchaliew P, Buathong N. Digital screen time usage, prevalence of excessive digital screen time, and its association with mental health, sleep quality, and academic performance among Southern University students. Front Psychiatry. 2025;16:1535631. doi: 10.3389/fpsyt.2025.1535631 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.AlKhenaizi AK, Shakeeb FN, Fredericks S, Gaynor D. Physical activity levels, recreational screen time, sleep quality and mood among young adult healthcare students at an international university in Bahrain: a cross-sectional study. BMJ Open. 2025;15(5):e093655. doi: 10.1136/bmjopen-2024-093655 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Moussa-Chamari I, Farooq A, Romdhani M, et al. The relationship between quality of life, sleep quality, mental health, and physical activity in an international sample of college students: a structural equation modeling approach. Front Public Health. 2024;12:1397924. doi: 10.3389/fpubh.2024.1397924 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mahin MI, Rahman MS, Rahman SM, et al. Factors impacting university students’ quality of life. PLoS One. 2025;20(8):e0329851. doi: 10.1371/journal.pone.0329851 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.The WHOQOL Group. The development of the World Health Organization quality of life assessment instrument (the WHOQOL). In: Orley J, Kuyken W, editors. Quality of Life Assessment: International Perspectives. Berlin, Germany:Springer;1994:41–57. doi: 10.1007/978-3-642-79123-9_4 [DOI] [Google Scholar]
- 22.Yao G, Chung CW, Yu CF, Wang JD. Development and verification of validity and reliability of the WHOQOL-BREF Taiwan version. J Formos Med Assoc. 2002;101(5):342–351. [PubMed] [Google Scholar]
- 23.Radloff LS. The CES-D Scale: a self-report depression scale for research in the general population. Appl Psychol Meas. 1977;1(3):385–401. doi: 10.1177/014662167700100306 [DOI] [Google Scholar]
- 24.Jiang L, Wang Y, Zhang Y, et al. The reliability and validity of the Center for Epidemiologic Studies Depression Scale (CES-D) for Chinese university students. Front Psychiatry. 2019;10:315. doi: 10.3389/fpsyt.2019.00315 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ribeiro ÍJS, Pereira R, Freire IV, de Oliveira BG, Casotti CA, Boery EN. Stress and quality of life among university students: a systematic literature review. Health Prof Educ. 2018;4(2):70–77. doi: 10.1016/j.hpe.2017.03.002 [DOI] [Google Scholar]
- 26.Wu HP, Lan HY, Tsai YL, et al. Gender as a moderator of the relationships among body composition, physical activity, basal metabolic rate, and BMI in Taiwanese university students. BMC Public Health. 2026;26(1):1222. doi: 10.1186/s12889-026-26810-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Tsai PS, Wang SY, Wang MY, et al. Psychometric evaluation of the Chinese version of the Pittsburgh Sleep Quality Index (CPSQI) in primary insomnia and control subjects. Qual Life Res. 2005;14(8):1943–1952. doi: 10.1007/s11136-005-4346-x [DOI] [PubMed] [Google Scholar]
- 28.Chien CP, Cheng TA. Depression in Taiwan: epidemiological survey utilizing CES-D. Seishin Shinkeigaku Zasshi. 1985;87(5):335–338. [PubMed] [Google Scholar]
- 29.Ringle CM, Wende S, Becker JM. SmartPLS 4. Bönningstedt: SmartPLS; 2024. Available from: https://www.smartpls.com. Accessed February 4, 2026. [Google Scholar]
- 30.Hair JF, Sarstedt M, Ringle CM, Gudergan SP. Advanced Issues in Partial Least Squares Structural Equation Modeling (PLS-SEM). 2nd ed. Thousand Oaks, CA: Sage; 2024. [Google Scholar]
- 31.Hair J, Hult GTM, Ringle C, Sarstedt M. A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). 3rd ed. Thousand Oaks, CA: Sage; 2022. [Google Scholar]
- 32.Henseler J, Ringle CM, Sarstedt M. Testing measurement invariance of composites using partial least squares. Int Mark Rev. 2016;33(3):405–431. doi: 10.1108/imr-09-2014-0304 [DOI] [Google Scholar]
- 33.Faul F, Erdfelder E, Buchner A, Lang AG. Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behav Res Methods. 2009;41(4):1149–1160. doi: 10.3758/BRM.41.4.1149 [DOI] [PubMed] [Google Scholar]
- 34.Kock N, Hadaya P. Minimum sample size estimation in PLS-SEM: the inverse square root and gamma-exponential methods. Inf Syst J. 2018;28(1):227–261. doi: 10.1111/isj.12131 [DOI] [Google Scholar]
- 35.Zhao X, Lynch JG Jr, Chen Q. Reconsidering Baron and Kenny: myths and truths about mediation analysis. J Consum Res. 2010;37(2):197–206. doi: 10.1086/651257 [DOI] [Google Scholar]
- 36.Alamir YA, Zullig KJ, Kristjansson AL, Wen S, Misra R, Montgomery-Downs H. A theoretical model of college students’ sleep quality and health-related quality of life. J Behav Med. 2022;45(6):925–934. doi: 10.1007/s10865-022-00348-9 [DOI] [PubMed] [Google Scholar]
- 37.Yan Y, Ye Q, Li T, Yuan Y, Dong J. Depression and mental health in late adolescents: the sequential mediating roles of quality of life and sleep regularity. BMC Psychiatry. 2026;26(1):316. doi: 10.1186/s12888-026-07944-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Lazarus RS, Folkman S. Stress, Appraisal, and Coping. New York: Springer; 1984. [Google Scholar]
- 39.Cummins RA. Subjective wellbeing, homeostatically protected mood and depression: a synthesis. J Happiness Stud. 2010;11(1):1–17. doi: 10.1007/s10902-009-9167-0 [DOI] [Google Scholar]
- 40.Cummins RA, Wooden M. Personal resilience in times of crisis: the implications of SWB homeostasis and set-points. J Happiness Stud. 2014;15(1):223–235. doi: 10.1007/s10902-013-9481-4 [DOI] [Google Scholar]
- 41.Su P, He M. Relationship between sleep quality and subjective well-being: resilience as a mediator and belief in a just world as a moderator. Front Psychiatry. 2023;14:1297256. doi: 10.3389/fpsyt.2023.1297256 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Lenneis A, Das-Friebel A, Tang NKY, et al. The influence of sleep on subjective well-being: an experience sampling study. Emotion. 2024;24(2):451–464. doi: 10.1037/emo0001268 [DOI] [PubMed] [Google Scholar]
- 43.Schandelmaier S, Guyatt G. Same old challenges in subgroup analysis—should we do more about methods implementation? JAMA Netw Open. 2024;7(3):e243339. doi: 10.1001/jamanetworkopen.2024.3339 [DOI] [PubMed] [Google Scholar]
- 44.Fingerhut A, Uranues S, Dziri C, et al. Interaction analysis of subgroup effects in randomized trials: the essential methodological points. Sci Rep. 2024;14(1):12619. doi: 10.1038/s41598-024-62896-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Han X, Zhou E, Liu D. Electronic media use and sleep quality: updated systematic review and meta-analysis. J Med Internet Res. 2024;26:e48356. doi: 10.2196/48356 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Haddaouy AE, Blaak H, Elkhalladi J, et al. Mediating role of sleep quality in the relationship between smartphone addiction and academic performance among nursing students in Morocco: a cross-sectional study. BMC Nurs. 2025;24(1):1281. doi: 10.1186/s12912-025-03949-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Pieh C, Humer E, Hoenigl A, et al. Smartphone screen time reduction improves mental health: a randomized controlled trial. BMC Med. 2025;23(1):107. doi: 10.1186/s12916-025-03944-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Shadzi MR, Rahmanian M, Heydari A, Salehi A. Structural validity of the Pittsburgh sleep quality index among medical students in Iran. Sci Rep. 2024;14(1):1538. doi: 10.1038/s41598-024-51379-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Martiadis V, Pessina E, Martini A, et al. Insomnia severity in psychiatric outpatients: real-world insomnia severity index data from an Italian Community Mental Health Center. Brain Sci. 2026;16(6):617. doi: 10.3390/brainsci16060617 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Cole RJ, Kripke DF, Gruen W, Mullaney DJ, Gillin JC. Automatic sleep/wake identification from wrist activity. Sleep. 1992;15(5):461–469. doi: 10.1093/sleep/15.5.461 [DOI] [PubMed] [Google Scholar]
- 51.Rundo JV, Downey R. Chapter 25—polysomnography. In: Levin KH, Chauvel P, editors. Handbook of Clinical Neurology. Amsterdam, Netherlands: Elsevier; 2019:381–392. [DOI] [PubMed] [Google Scholar]
- 52.Aydin F, Aydin A. Relationship among sleep quality, quality of life and academic self-efficacy of university students. Curr Psychol. 2024;43(24):21110–21119. doi: 10.1007/s12144-024-05929-2 [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and/or analysed during the current study are not publicly available in order to protect the privacy of the study participants, but are available from the corresponding author on reasonable request.
