Academic stress and online learning engagement in medical students: the parallel mediating roles of sleep quality and positive academic emotions

Abstract

Background

Academic stress is a significant factor affecting students’ mental and physical health. There exists a complex relationship between medical students’ academic stress and their learning engagement in online learning environments. While previous studies have explored the various consequences of academic stress, the way in which it influences learning engagement through sleep quality and positive academic emotions remains underexplored. This study aimed to elucidate the pathways linking academic stress to learning engagement in online environments and to examine the mediating roles of sleep quality and positive academic emotions.

Method

A cross-sectional design was employed, with 400 medical undergraduates completing validated questionnaires assessing academic stress, sleep quality, positive academic emotions, and learning engagement. Data were collected using the Perception of Academic Stress, the Athens Insomnia Scale, the Academic Emotions Questionnaire, and the Student Engagement Scale. Structural equation modeling was conducted to test hypothesized associations. In this study, the common method bias test, descriptive statistics, correlation analysis, mediation effect analysis and moderated effect analysis were conducted using SPSS 23.0, PROCESS plug-in. Since the data of the four scales in this study came from the same subjects, the Harman one-way test was used to test the common method bias.

Result

Academic stress directly and negatively predicted learning engagement. Independent mediation effects were observed through sleep quality and positive academic emotions. However, the hypothesized chain mediation via sleep quality → positive academic emotions was non-significant.

Conclusion

Contrary to prior assumptions, academic stress impacts learning engagement through parallel rather than chained mediation pathways. This challenges traditional mediation models and underscores the need for dual-targeted interventions addressing both sleep hygiene and emotional regulation in online medical education.

Clinical trial number

Not applicable.

Introduction

The World Health Organization declared the COVID-19 outbreak a pandemic in March 2020, accelerating the integration of online courses into students’ lives [1]. In medical education, advancements in information technology and pandemic-related constraints significantly increased the prevalence of online learning for medical students [2, 3]. While online modalities address certain limitations of traditional classroom instruction, such as spatial constraints and scheduling inflexibility [4, 5], their effectiveness in medical training remains debated due to inadequate supervision and inconsistent process evaluation [6, 7]. A critical challenge lies in sustaining learning engagement, a multidimensional construct encompassing cognitive, behavioral, and affective investment in academic tasks [8–10]. One of the most important factors in online education is learning engagement [11]. However, learning engagement in online medical education is often suboptimal, creating an urgent need to identify modifiable factors that can enhance it [10, 12].

Medical students undergo proportionately undue stress relative to other students [13], an issue which is similarly well-documented among nursing students in a range of learning environments, including clinical practice [14]. This is particularly pronounced in the Chinese context. In China, clinical medical education begins after high school graduation with students gaining admission to medical schools based on their National College Entrance Examination scores. They undergo a five-year undergraduate program without specialization, followed by a three-year standardized residency training at the master’s level after passing the National Postgraduate Entrance Examination. This rigorous path requires managing a heavy workload of coursework and clinical hours, with significant competitive pressure due to the high demand for popular specialties [15]. This can easily lead to academic stress and affect their learning engagement and emotional experience [16]. Unlike in Western countries, the excessive focus on academic performance by families and individuals has significantly intensified the academic stress on medical students [17]. Such stress is frequently associated with negative emotional responses toward academic tasks and lower learning engagement [18]. Moreover, sleep quality and positive academic emotions may play a mediating role between academic stress and learning engagement. Medical students are recognized as a population that is particularly vulnerable to sleep-related problems and depression [19–21]. Within the demanding framework of Chinese medical education, where heavy workloads and future career prospects are closely tied to academic performance [15], the ability to maintain good sleep and positive academic emotions becomes even more critical. Evidence indicates that medical students’ learning engagement is significantly affected by sleep quality and positive academic emotions [22]. Recent global research has also emphasized the role of stress management and emotional regulation in sustaining learning engagement among health professions students. For instance, Ayed and Amoudi reported that stress during clinical practice significantly influenced coping behaviors among physical therapy students, highlighting the importance of supportive learning environments [23]. Similarly, Toqan et al. found that interventions such as progressive muscle relaxation effectively reduced anxiety and improved students’ emotional readiness for clinical learning [24]. These studies underscore the importance of academic stress and positive academic emotions management across different health professions education contexts. Therefore, understanding how sleep quality and positive academic emotions play a role in the relationship between academic stress and learning engagement is critical to improving medical students’ learning engagement.

Prior research predominantly examined the direct effects of academic stress on positive academic emotions, sleep quality, and learning engagement. Moreover, no integrated analysis exists on how academic stress concurrently relates to learning engagement through the potential mediators of sleep quality and positive academic emotions in online learning environments. Crucially, most studies originated from education and social sciences disciplines, limiting their applicability to medical education. Medical students typically experience higher academic stress, lower positive emotional states, and poorer sleep quality than students in other fields, necessitating domain-specific research [25–28]. This study addresses these gaps in the field of medical education by:

1. Examining the direct effect of academic stress on online learning engagement.

2. Quantifying the mediating roles of sleep quality and positive academic emotions.

3. Proposing evidence-based interventions to optimize engagement in medical online learning.

The effect of academic stress on learning engagement

Academic stress refers to the psychological and emotional stress that students experience during their academic pursuits [12]. Learning engagement denotes the positive, diligent, and focused state demonstrated by students during educational activities [29].

Academic stress demonstrates a significant negative association with learning engagement. The Conservation of Resources Theory provides a mechanistic framework for this relationship: when students encounter academic stress, they deplete psychological resources (e.g., emotional regulation, cognitive capacity, social support), which reduces their commitment to learning [30, 31]. Recent studies suggest a strong negative correlation between academic stress and engagement levels [12, 18, 32, 33].

For this study, we employed a learning engagement scale grounded in the well-established tripartite framework (behavioral, emotional, cognitive) [34], which provides a comprehensive and validated measure of core engagement processes that are fundamental across learning environments. While newer, context-specific scales such as the Online Student Engagement Scale offer valuable insights into unique digital domains [35], our chosen framework was deemed most appropriate for capturing the universal mechanisms of engagement that underpin our hypotheses regarding stress and mediators in the Chinese medical student population. In this study, academic stress was assessed using an adapted version of the Perception of Academic Stress Scale, a concise and widely used instrument that effectively captures students’ subjective experience of stress stemming from academic demands [36].

The mediating role of positive academic emotions

Positive academic emotions refer to positive affective experiences, such as pleasure, curiosity, and accomplishment (e.g., joy, hope, pride). These emotions enhance motivation and promote learning engagement [37]. The “positive/negative” labels refer to valence (pleasant vs. unpleasant) rather than inherent desirability or universal effects. The functional impact of emotions depends on context, appraisal, and regulation. Positive academic emotions have been shown to broaden an individual’s horizon, enhance resilience, and broaden the scope of one’s attention [22]. This helps students cope better with academic stress to maintain a higher level of learning engagement.

Academic stress is one of the most important factors affecting students’ emotions. Higher stress is associated with a diminished capacity to retain interest and enthusiasm, as well as lower learning engagement [38]. Current studies suggest a strong negative correlation between academic stress and positive academic emotions [39–41].

Current research suggests that positive academic emotions significantly predict students’ learning engagement [42–45]. The Broaden-and-Build Theory posited that positive academic emotions expanded cognitive flexibility, allowing individuals to perceive broader opportunities and enhance problem-solving capabilities [46]. This cognitive expansion enabled students to adopt open-minded approaches to academic stress, explore innovative learning strategies, and sustain learning engagement under high-pressure conditions [47]. Positive academic emotions not only help to increase students’ emotional engagement but also increase their cognitive engagement and behavioral engagement and promote their concentration and persistence in learning activities [41]. Consistent with this, a study found that positive academic emotions were positively associated with higher levels of learning engagement [48]. Zhao et al. found that positive academic emotions and learning engagement showed a significant positive correlation [45].

We measured these positive academic emotions using selected items from the Academic Emotions Questionnaire, a scale designed to capture the various emotions students encounter in academic settings [49].

The mediating role of sleep quality

Empirical evidence indicates that academic stress significantly impacts students’ psychophysiological functioning, with elevated stress levels correlating negatively with sleep quality [50]. For example, a study of medical students demonstrated that those experiencing higher stress reported poorer sleep quality and more insomnia symptoms [51]. The stress-diathesis model proposed by Spielman and Perlis identifies stressors as a major driver in the development of chronic insomnia. Mechanistically, academic stress activates the hypothalamic-pituitary-adrenal (HPA) axis, inducing cortisol secretion irregularities that perpetuate circadian rhythm disruptions [52]. At the same time, academic stress is often linked to lower sleep quality, as it correlates with students’ emotional state and physiological responses [53]. Stressed students often face problems such as insomnia or light sleep, which affects the overall quality of sleep [54]. Recent studies suggest that academic stress has a significant negative impact on sleep quality [26, 55–57].

High-quality sleep improves attention span and cognitive performance, thereby boosting learning engagement. Conversely, sleep deprivation or poor sleep quality reduces students’ energy for learning, impairing concentration, memory retention, and learning engagement [58]. It has been shown that poorer sleep quality leads to poor concentration and memory loss, which significantly reduces learning engagement [59]. Meanwhile, a study of college students directly demonstrated that poor sleep quality predicted lower learning engagement among university students, highlighting sleep’s crucial role in sustaining learning engagement [60]. Thus, the established link between poor sleep and diminished learning engagement provides a rationale for its role as a mediator in the relationship between academic stress and learning engagement [61–63].

The standard consolidation theory and the Reward Activation Model suggest that sleep plays a role in activating the emotional and reward systems, as well as consolidating memories that hold high emotional or motivational significance [64, 65]. Specifically, the standard consolidation theory posits that sleep is essential for processing and stabilizing emotional memories. Complementing this framework, the Reward Activation Model provides a specific neurobiological mechanism by highlighting the activation of the mesolimbic dopaminergic (ML-DA) reward system during sleep, particularly during REM sleep. This activation is hypothesized to reactivate reward-related or emotionally salient memories, thereby facilitating their selective consolidation and influencing the motivational content of dreams. Therefore, integrating these two theoretical perspectives, it is plausible that poorer sleep quality, by disrupting the ML-DA-mediated consolidation of positive emotional experiences, is associated with reduced positive academic emotions. Sleep disorders, deprivation, and disruption can impair motivation and emotional functioning [66]. Shen et al. also found that shorter sleep duration was specifically associated with lower positive academic emotions [67]. Consequently, poorer sleep quality is associated with reduced positive academic emotions [68]. Sleep quality and positive academic emotions may constitute a chain mediating effect of academic stress on learning engagement.

To evaluate sleep quality, the sleep quality of participants was measured using a culturally adapted version of the Athens Insomnia Scale, a self-report tool validated for screening insomnia-related symptoms and assessing nocturnal sleep quality [69].

Conceptual framework and hypotheses

The proposed conceptual model is grounded in an integration of the Conservation of Resources Theory, the stress-diathesis model, the standard consolidation model, the Reward Activation Model, and the Broaden-and-Build Theory. The direct negative effect of academic stress on learning engagement is primarily explained by the Conservation of Resources Theory, which posits that academic stress depletes the psychological resources necessary for sustained learning engagement [31]. Furthermore, the model specifies two parallel mediating pathways. The pathway through sleep quality is informed by the stress-diathesis model and related psychophysiological research, which indicates that academic stress activates physiological systems that disrupt circadian rhythms and impair sleep [52, 53]. Concurrently, the pathway through positive academic emotions is informed by the Broaden-and-Build Theory, which suggests that positive academic emotions expand cognitive flexibility, allowing individuals to perceive broader opportunities and enhance problem-solving capabilities [46]. This cognitive expansion enabled students to adopt open-minded approaches to academic stress, explore innovative learning strategies, and sustain learning engagement under high-pressure conditions [47]. In addition to these parallel pathways, a chain-mediating effect is theorized. Drawing on the standard consolidation model and the Reward Activation Model, impaired sleep is anticipated to further disrupt emotional and reward system functioning, thereby hindering the experience of positive academic emotions [64, 65]. In short, academic stress disrupts sleep, and this physiological/restorative dysfunction, in turn, directly undermines the ability to develop positive academic emotions (e.g., curiosity, pride, enjoyment), which ultimately impacts learning engagement. Consequently, this integrated theoretical framework posits that academic stress impacts learning engagement both directly through resource depletion and indirectly through the independent mediating roles of sleep quality and positive academic emotions, as well as through their chain-mediating role (Fig. 1).

Fig. 1.

Conceptual framework

The purpose of this study is to examine the relationship between academic stress and learning engagement in an online learning context. Additionally, the study examines the mediating role of sleep quality and positive academic emotions in the relationship between academic stress and learning engagement. The key point is that our model not only assesses the independent effects of these mediating variables but, more importantly, examines the possibility of parallel mediating pathways. Based on relevant theories and existing studies, this study proposes the following hypotheses:

• H1: Academic stress has a direct effect on learning engagement.

• H2: Sleep quality mediates the relationship between academic stress and learning engagement.

• H3: Positive academic emotions mediate the relationship between academic stress and learning engagement.

• H4: Sleep quality and positive academic emotions play a chain-mediated effect on the relationship between academic stress and learning engagement.

Methods

Research context and participants

This study was conducted at a public medical university in eastern China. The university has adopted a blended teaching model that combines traditional classroom and online learning formats to provide students with more flexible learning options. The university provides students with a variety of learning resources and platforms, including MOOC, Wisdom Tree, Yangtze River Rain Classroom, and Study Pass. Students can usually learn by watching course videos, accessing course materials and completing relevant tests.

This study employed a convenience sampling method to administer questionnaires to undergraduate students at a public university in eastern China. Recruitment took place over one week in November 2024. We invited undergraduate students from different grades and majors to participate in the research. To be included in the study, participants had to meet the following criteria: (1) be an undergraduate student at the university, (2) have taken at least one online course in the current semester, and (3) voluntarily agree to participate by starting the online questionnaire. Participants were required to complete electronic questionnaires on academic stress, sleep quality, positive academic emotions, and learning engagement via an electronic questionnaire platform (Wenjuanxing). A total of 436 questionnaires were collected initially. After applying data cleaning procedures, which excluded responses submitted too quickly or exhibiting clear patterns of random answering (e.g., straight-lining), 400 valid questionnaires were retained, resulting in an effective response rate of 91.7%. The mean age of the participants was 18.84 years (SD = 1.04). Among them, 173 (43%) were male and 227 (57%) were female. In terms of grade distribution, 196 (49.00%) were freshmen, 159 (39.80%) were sophomores, 33 (8.30%) were juniors, and 12 (3.00%) were seniors. The sample exhibits a relatively balanced distribution across gender and specialization, broadly reflecting the demographic characteristics of the undergraduate medical student population at this institution. The study was approved by the Ethics Committee of Central China Normal University(CCNU-IRB-202409002 A). The participants were informed that their information would be kept confidential, and all participants filled out an informed consent form before starting.

Measures

Four scales were employed to measure the key variables: Academic Stress, Sleep Quality, Positive Academic Emotions, and Learning Engagement. All scales used in the study were readily available online and were originally developed in the English language. To ensure their suitability for Chinese students, we followed a rigorous translation and adaptation process. First, the original English scales were translated into Chinese by two bilingual experts. Then, another bilingual expert, who had not seen the original versions, translated the Chinese drafts back into English. The back-translated versions were compared with the originals, and any discrepancies were discussed and resolved by the research team to finalize the Chinese versions. All items except those on the sleep quality scale are measured using a five-point Likert scale, while the remaining items employ a descriptive format. Participants were asked to indicate their agreement with the statements on a scale of 1 (strongly disagree) to 5 (strongly agree).

Face validity assessment would have ensured the suitability of these measures for the selected population [70]. To establish face validity, the Chinese version of the questionnaire was pilot-tested with 30 medical students who were also taking online courses. Participants were asked to evaluate the clarity, comprehensibility, and relevance of each item to ensure the questions were appropriately interpreted within the local context. Based on their feedback, we made minor wording adjustments to improve understanding while maintaining the original constructs’ integrity.

Academic stress scale

The Academic Stress Scale is adapted from the Perception of Academic Stress [36]. The scale consists of four items, such as “I can make academic decisions easily when studying online” and “I have enough time to relax after completing my study tasks.” The reliability coefficient (Cronbach’s α) of the scale was 0.812.

Sleep quality scale

The Sleep Quality Scale is adapted from the Athens Insomnia Scale [69]. The scale consists of four items, such as “time to fall asleep” and “nocturnal awakenings”. To reduce cognitive load on participants, we retained the original scoring model but rewrote the items as self-report statements. For example, the item assessing sleep latency was presented as: “I have difficulty falling asleep at night,” with response options corresponding to the original AIS scoring: 1 = “No problem (fall asleep immediately after turning off the light),” 2 = “Slightly delayed (take some time to fall asleep after turning off the light),” 3 = “Obviously delayed (take considerable time to fall asleep),” 4 = “Very delayed (take very long time to fall asleep),” and 5 = “Extremely delayed or no sleep at all (completely insomnia after turning off the light).” This adaptation approach has been successfully implemented in previous Chinese student populations, ensuring adequate face validity for the target population. The scale’s construct validity was supported by confirmatory factor analysis in our study, with all factor loadings exceeding 0.5. The reliability coefficient (Cronbach’s α) of the scale was 0.723.

Positive academic emotions scale

The Positive Academic Emotions Scale is adapted from the Academic Emotions Questionnaire [49]. The scale consists of three items, such as “I enjoy learning new things through online platforms” and “Completing online learning tasks makes me feel proud.” The reliability coefficient (Cronbach’s α) of the scale was 0.771.

Learning engagement scale

The Learning Engagement Scale was adapted from a 5-point Likert-scaled instrument developed by Fredricks et al., measuring three types of engagement: behavioural engagement, emotional engagement, and cognitive engagement [29, 71]. This scale was adapted by Sun and Rueda into a 19-item version tailored for the online learning environment of undergraduate students, with both validity and reliability analyses completed [34]. We selected and adapted 8 items that best captured these three dimensions. Sample items include “I follow the rules of the online course” (behavioral), “I think online learning is fun” (emotional), and “When I encounter difficulties while learning online, I find ways to figure them out” (cognitive). The wording of some items was slightly modified to reference the “online” learning environment specifically. The reliability coefficient (Cronbach’s α) of the scale was 0.865.

Data analysis

In this study, the common method bias test, descriptive statistics, correlation analysis, mediation effect analysis and moderated effect analysis were conducted using SPSS 23.0, PROCESS plug-in. Since the data of the four scales in this study came from the same subjects, the Harman one-way test was used to test the common method bias. The results showed that there were a total of four factors with eigenvalues greater than 1, and the variance explained by the first factor was 38.158%, which was less than the critical criterion of 40%, indicating that the data in this study did not have a serious problem of common method bias.

Before analyzing the data, we verified all responses. We implemented the following data cleaning procedures: (1) responses with a completion time of less than 5 min were considered careless and removed; (2) responses selecting only one option (e.g., all selections as 1 or all as 5) were discarded; and (3) responses with missing values on key demographic or scale items were discarded. This process was to ensure the quality and seriousness of the data. First, the data were tested for normality. After conforming to normal distribution, the data were analyzed with descriptive statistics on demographic variables to explore the distributional characteristics of the samples. Then, correlation analyses were conducted to test whether there was a correlation between sleep quality, academic stress, learning engagement, and positive academic emotions. Last, the present study used 5,000 repetitions of the sampling method, and the confidence level was set to 95%, selecting the Model 4 (mediation model) to test the mediation effect.

Results

Measurement model

In this study, a confirmatory factor analysis (CFA) was conducted to assess the construct validity of the scale. Convergent validity was established by testing the reliability of each measure as well as the composite reliability (CR) and average variance extracted (AVE) [72], as shown in Table 1. Each standardized factor loading was greater than 0.400, ranging from 0.441 to 0.917, indicating that the applicability of each factor loading met the requirements. According to Fornell and Larcker [73], an average variance extracted (AVE) of 0.4 is acceptable when the composite reliability (CR) of the structure is high. In this study, an AVE of 0.457 is acceptable. Because the CR values for all measured dimensions were greater than 0.7 (see Table 1), this proves the convergent validity and reliability of the measurement model. In general, the CR value should be greater than 0.7, and the CR value in this study performed well (see Table 1). In addition, χ2/df < 5.0, RMSEA < 0.1, RMR < 0.05,SRMR < 0.1, which indicates a good model fit [74].

Table 1.

Reliability and convergent validity of the model

Factors	Items	Std. Factor Loading☐	SMC	CR	AVE
PAE	PAE1	0.859	0.738	0.792	0.568
	PAE2	0.823	0.678
	PAE3	0.536	0.288
LE	LE1	0.441	0.194	0.866	0.457
	LE2	0.583	0.340
	LE3	0.754	0.569
	LE4	0.853	0.727
	LE5	0.853	0.727
	LE6	0.600	0.360
	LE7	0.668	0.446
	LE8	0.544	0.296
AS	AS1	0.647	0.419	0.800	0.502
	AS2	0.724	0.524
	AS3	0.783	0.614
	AS4	0.671	0.450
SQ	SQ1	0.522	0.273	0.752	0.516
	SQ2	0.659	0.434
	SQ3	0.917	0.840

Descriptive statistics and correlations

Correlation analyses showed (see Table 2) that academic stress was significantly negatively correlated with sleep quality (p < 0.001, r = −0.244); learning engagement was significantly positively correlated with sleep quality (p < 0.001, r = 0.231); learning engagement was significantly negatively correlated with academic stress (p < 0.001, r = −0.622); positive academic emotions was significantly positively correlated with learning engagement (p < 0.001, r = 0.689); positive academic emotions and academic stress (p < 0.001, r = −0.546) were significantly negatively correlated; positive academic emotions were significantly positively correlated with sleep quality (p < 0.01, r = 0.142).

Table 2.

Descriptive statistics and pearson’s correlations (N = 400)

	M	SD	1	2	3	4
1.Sleep quality	4.001	0.636	1.00
2.Academic stress	3.728	0.649	−0.244***	1.00
3.Learning engagement	3.708	0.564	0.231***	−0.622***	1.00
4. Positive academic emotions	3.825	0.649	0.142**	−0.546***	0.689***	1.00

*p < 0.05, **p < 0.01, ***p < 0.001

The mediation model

As shown in Table 3, academic stress was a significant direct negative predictor of learning engagement (β = −0.288, p < 0.001); after the inclusion of sleep quality, academic stress became a significant positive predictor of learning engagement (β = 0.071, p = 0.02), indicating that sleep quality mediated the relationship between academic stress and learning engagement. In addition, after the inclusion of positive academic emotions, academic stress also became a significant positive predictor of learning engagement (β = 0.431, p < 0.001), suggesting that positive academic emotions also mediated the relationship between academic stress and learning engagement. Moreover, academic stress significantly negatively predicted sleep quality (β = −0.240, p < 0.001) and significantly negatively predicted positive academic emotions (β = −0.544, p < 0.001). In contrast, the path from sleep quality to positive academic emotions was not significant (β = 0.009, p = 0.836). Figure 2 reflects the regression coefficients, 95% bootstrap CIs, and significance levels more clearly for each path.

Table 3.

Test for the chain mediation model

Regression equation		Overall fit indices			Significance
Outcome variables	Predictive variables	R2	F	SE	β☐	t
Sleep quality	Academic stress	0.057	25.294***	0.048	−0.240	−5.029***
Positive academic emotions	Academic stress	0.295	84.399***	0.043	−0.544	−12.546***
	Sleep quality				0.009	0.207
Learning engagement	Academic stress	0.563	172.569***	0.035	−0.288	−8.229***
	Sleep quality				0.071	2.333*
	Positive academic emotions				0.431	12.564***

***p < 0.001, **p < 0.01, *p < 0.05

Fig. 2.

The Chain Mediation Model. N = 400, ***p < 0.001, **p < 0.01, *p < 0.05

As shown in Table 4, the direct effect of academic stress on learning engagement was − 0.288 (Boot 95% CI [−0.357,−0.220]); Boot 95% confidence interval excluding 0, indicating that the direct effect of academic stress on learning engagement was significant and hypothesis 1 was supported. The total indirect effect was − 0.253 with 95% confidence interval excluding 0. Among them, the mediating effect of sleep quality was − 0.017 (Boot95% CI [−0.040,−0.002]); Boot95% confidence interval excluding 0. Therefore, hypothesis 2 was supported. The mediating effect of positive academic emotions was − 0.235 (Boot95% CI [−0.336,−0.210]); the Boot95% confidence interval did not include 0; therefore, hypothesis 3 was also supported. However, the chain-mediated effect of academic stress on learning engagement through sleep quality and positive academic emotions was − 0.001 (Boot95% CI [−0.013,0.010]), therefore, Hypothesis 4 could not be tested and was not supported. The results could not show a significant chain-mediated effect.

Table 4.

The direct and indirect effect of chain mediation model

Effect types	Effect	Boot SE	Boot 95%CI
Direct effect	−0.288	0.035	[−0.357,−0.220]
Total indirect effect	−0.253	0.033	[−0.358,−0.228]
AS→SQ→LE	−0.017	0.010	[−0.040,−0.002]
AS→PAE→LE	−0.235	0.032	[−0.336,−0.210]
AS→SQ→PAE→LE	−0.001	0.006	[−0.013,0.010]

Note: AS Academic stress, SQ Sleep quality, PAE Positive academic emotions, LE Learning the engagement

Discussion

The direct effect of academic stress on learning engagement

Our results confirm that academic stress has a significant direct negative effect on learning engagement, supporting Hypothesis 1. This finding aligns with the study by Alkharj et al. [18], which demonstrated a significant negative correlation between learning engagement and academic stress among Saudi undergraduate nursing students. This relationship is particularly critical in medical education, where learning engagement is a crucial key component for developing future clinical competence and sustaining work engagement [75]. The distinct nature of medical training introduces distinctive, profession-specific stressors. These stressors encompass the immense volume and complexity of foundational knowledge, high-stakes licensure examinations, and anxieties concerning future clinical responsibilities and patient care [13, 76]。 Such stressors negatively impact learning engagement. The Conservation of Resources Theory (COR) provides a robust framework for understanding this process [31]. According to Conservation of Resources Theory (COR) [31], various stressors in the academic environment—such as exams, assignments, time management challenges, social relationships, and uncertainty about future careers—deplete students’ psychological resources, including emotional regulation, cognitive capacity, and social support. This resource depletion is especially detrimental for medical students. Mastering a vast curriculum and managing the emotional demands of clinical training are fundamental to their professional development. When excessive academic stress drains these resources, it directly compromises the cognitive and affective engagement essential for developing clinical competence. This not only diminishes learning efficiency but can also trigger emotional exhaustion and reduced motivation, further undermining learning engagement [30]. Additionally, numerous studies have examined the impact of academic stress on learning engagement. For instance, Aherne and van both found that academic stress is a critical factor contributing to decreased learning engagement [32, 33]. Notably, while Alkharj focused on Saudi nursing students, our study was conducted in a different cultural context, yet both reached a similar conclusion: academic stress negatively impacts students’ engagement in learning [18]. This suggests that the adverse effects of academic stress on learning engagement are cross-culturally generalizable and not confined to a specific culture or subject area. Although our investigation took place in the online learning environment, the challenges of sustaining learning engagement under stress also extend to clinical settings [77]. Therefore, mitigating academic stress requires concrete actions. For instance, in online learning environments, instructors can help medical students learn from clinical cases by giving specific feedback. This means praising the parts the medical student handled correctly and then giving them clear, straightforward steps to correct and improve the parts they did not. This approach helps students manage the uncertainty of complex medical knowledge and reduces the academic stress associated with subjective evaluation. Instructors can scaffold complex topics by providing pre-session concept maps and post-session summary infographics, which reduce the cognitive load of organizing vast information. These targeted interventions help preserve psychological resources by addressing key academic stressors, thereby fostering conditions that support sustained learning engagement and professional growth.

The mediating effect of sleep quality and positive academic emotions

The statistically mediated pathway through sleep quality indicates that academic stress exerts a deleterious influence on learning engagement via its detrimental effects on nocturnal rest patterns. The results of the study show that medical students with higher academic stress tend to have poorer sleep quality. Such sleep problems can be caused by anxiety and worry caused by high stress, misalignment of the body clock, and increased levels of stress hormones in the body. Previous studies and the stress-diathesis model have shown a significant association between stress and sleep quality in medical students [78]. Daily stress may increase students’ mental activity before falling asleep, causing them to subjectively experience more thought fluctuations and further prolong their perceived time to fall asleep, ultimately affecting their sleep quality [79]. This is particularly relevant in medical education, where demanding coursework, exams, and clinical rotations create intense academic stress that significantly disrupts sleep [15]. According to Taylor, among the multiple variables examined, sleep duration is the single most significant predictor of student GPA [80]. This finding highlights the importance of adequate sleep for academic performance, with poor sleep quality not only affecting the academic performance of medical students but also reducing their learning engagement. Wu further revealed that there is a significant positive correlation between learning engagement and learning achievement [10]. In medical education, the implications extend beyond grades. Chronic sleep disruption may erode the learning engagement and cognitive stamina required for safe clinical practice, directly impacting future patient care [81, 82]. Therefore, medical schools should consider integrating sleep health promotion into student wellness programs to safeguard both academic and clinical competency.

The results of the study show that the impact of academic stress on learning engagement is not only reflected in the direct negative effects, but also indirectly through the increase of positive academic emotions. Broaden-and-Build Theory holds that positive emotions (such as happiness, curiosity, pride and satisfaction, etc.) can “broaden” individuals’ cognitive horizons, enabling them to discover more possibilities and opportunities, thus enhancing their ability to cope with challenges and creativity [46]. Within medical training, positive academic emotions may help learners cope with the academic stress arising from intensive coursework and clinical rotations through more flexible cognitive approaches [83]. This capacity is vital for medical students, who must continuously master complex knowledge and adapt to novel clinical situations. Positive academic emotions, such as curiosity about disease mechanisms or pride in mastering a clinical skill, can promote exploratory behaviors in adopting novel learning strategies and sustain learning engagement under the high-pressure conditions typical of medical training [47, 84] This finding is consistent with previous findings and is consistent with the fact that those students with positive academic emotions are more likely to devote more energy to their studies and thus achieve higher learning engagement in their studies [43, 45]. Specifically, positive affective states function as psychological resilience mechanisms that enhance medical students’ capacity to implement adaptive coping strategies when confronting academic stress [85]. Such affective regulatory processes enable learners to sustain cognitive engagement in academic pursuits [86]. These emotional resources are vital for long-term professional development, as they help build the resilience and professional identity needed to navigate the emotional demands of patient care and mitigate burnout risk in future clinical practice [87]. Therefore, we recommend that educator development programs incorporate training on fostering positive academic emotions. Such training should focus on practical strategies like providing strengths-based feedback and structuring learning tasks to create mastery experiences, which collectively support the development of a resilient and professionally engaged physician workforce.

The chain mediation effect of sleep quality and positive academic emotions

Contrary to Hypothesis 4, our results do not support a significant chain-mediating effect of sleep quality and positive academic emotions between academic stress and learning engagement. Combined with the low correlation between the two mediating variables (r = 0.142), this suggests that in medical student populations, sleep quality and positive academic emotions may influence the relationship between academic stress and learning engagement in parallel rather than through a chain mediation mechanism. This finding diverges from the direct predictions of the standard consolidation theory and the Reward Activation Model in this context [66], reflecting the multidimensional and complex nature of how academic stress affects learning engagement. This parallel rather than chain mediation pattern may be explained by the specific medical education culture in which these students are embedded. The highly competitive nature of Chinese medical education, where heavy academic burdens are closely tied to future career prospects [15], may fundamentally shape this parallel mediation. In such a high-stakes and high-pressure environment, sleep quality and positive academic emotions appear to impact learning engagement through distinct and relatively independent channels.

First, regarding sleep quality, our findings suggest its role as an independent mediator is largely cognitive. It should be noted that medical education has particularly high requirements, with a vast amount of knowledge content and complex problem-solving demands [13]. If sleep quality is poor, it will affect the ability to maintain concentration during long-term study and also hinder the consolidation of memory for complex medical concepts [58]. This will directly reduce learning engagement, but it does not necessarily affect positive academic emotions. This direct cognitive approach can explain why sleep quality can play an independent role in the relationship between academic stress and learning engagement.

In contrast, the mediating role of positive academic emotions appears to operate through different pathways. Within the highly competitive Chinese medical education system, where career prospects are closely tied to academic performance, students’ positive academic emotions are primarily driven by achievement-related experiences [15]. These include successfully mastering complex clinical skills [43], receiving recognition from faculty and peers [88], and making progress toward long-term professional goals. The excessive focus of families and individuals on academic performance further externalizes positive academic emotions [17], linking them closely to achievement milestones. Consequently, in competitive environments, students may develop psychological coping strategies. Separate physical fatigue from academic emotional experiences. In this way, even if they don’t get enough sleep, they can still maintain positive academic emotions. Immediate feedback on academic performance, such as encouragement from teachers or mastery of clinical skills, may provide students with strong emotional motivation, masking the subtle effects of sleep deprivation. This prioritization of academic striving over physiological needs could effectively decouple the experience of positive academic emotions from variations in sleep quality, leading to their function as parallel mediators. The parallel operation of these mediators suggests they represent distinct pathways through which academic stress affects learning engagement. These findings carry significant implications for medical education in China, where the highly competitive and demanding training environment often exacerbates both sleep problems and emotional challenges. Our results advocate for the development of targeted programs that address both pathways simultaneously. To improve sleep quality, medical schools could implement structured interventions such as time management training to reduce academic overload, cognitive-behavioral therapy for insomnia (CBT-I) delivered through digital platforms, and education on sleep hygiene that emphasizes the critical role of sleep in cognitive performance and emotional regulation. Concurrently, fostering positive academic emotions requires creating mastery-oriented learning experiences, providing strengths-based feedback, and cultivating supportive faculty-student interactions. This integrated, dual-pronged approach is essential for enhancing online learning engagement and, by extension, sustaining the long-term professional development of medical students. Future research should further explore the longitudinal dynamics of these parallel mediation pathways. Subsequent studies could also extend our model by incorporating institutional factors, such as faculty support and curriculum design quality, which are strong predictors of student success [89]. Combining the psychological supports identified in this study with broader institutional reforms might produce the most substantial improvements in student well-being and learning engagement.

Study limitations

This study has several limitations. Firstly, the cross-sectional design employed in this research restricts our capacity to establish causal relationships. Consequently, future studies should adopt experimental or longitudinal designs to validate the findings of this study. In particular, longitudinal SEM is very valuable for delimiting time priority and strengthening causal inference between research variables. Secondly, the insignificant chain mediation effect observed in this study may be partly attributed to the specific culture and educational background of Chinese medical education. Sleep quality and positive academic emotions seem to work independently. Future cross-cultural research should study these relationships in a diverse educational environment to determine the generalization we find. Thirdly, the sample for this study was limited to students from a specific major at a university in a particular region, which may result in a homogeneous sample group that lacks diversity in terms of student backgrounds and academic levels. To enhance the generalizability of the findings, future research should consider expanding the diversity of the sample and incorporating a broader range of variables to better represent a wider student population. The model developed in this study partially elucidates the relationships among academic stress, learning engagement, sleep quality, and positive academic emotions. However, it incorporates only a subset of potential factors and does not fully account for other significant variables that may influence online learning. For instance, factors such as students’ learning styles and subject-specific differences were not included, which limits the model’s explanatory power. Future research should explore the mechanisms through which these and other factors impact online learning. In addition, the use of brief, adapted versions of the scales, while practical, may have constrained the variance and measurement precision of the constructs. Furthermore, our measure of learning engagement was a general scale adapted for online use. Future studies could employ online-specific instruments, such as the Online Student Engagement (OSE) Scale, which was validated in an Italian context and offers a multi-dimensional assessment (e.g., Skills, Emotional, Participation, Performance). Finally, this study utilized self-report questionnaires to measure variables, a method that may introduce subjective bias and limit insights into students’ learning processes and experiences. This is particularly relevant for sleep quality, which was assessed subjectively and may not fully align with objective physiological measures. Participants’ responses could be influenced by social expectations or personal perceptions, potentially leading to inaccuracies in reflecting their true learning behaviors and sleep patterns. To address these limitations, future research could adopt a mixed-methods approach and incorporate objective measures, such as actigraphy, to more comprehensively assess the factors influencing learners’ online learning persistence and the underlying mechanisms of these effects.

Conclusion

Guided by relevant theories, our study found that academic stress influences learning engagement through parallel rather than chained mediation pathways. These findings emphasize sleep and emotional regulation as key intervention targets. Educators should mitigate workload-induced sleep deprivation by optimizing academic schedules and fostering supportive environments to enhance positive academic emotions. In conclusion, targeted interventions focusing on stress management, emotional well-being, and sleep hygiene are essential to enhance medical students’ learning engagement and learning outcomes in online education contexts.

Biographies

Li he

is an undergraduate student of Department of Clinical Skills Training Center, Zhujiang Hospital, Southern Medical University. His research interests include medical education.

Wei Li

is a staff member of Department of Clinical Skills Training Center, Zhujiang Hospital, Southern Medical University. His research interests include medical education, educational neuroscience and orthopedics.

Yueyi Zhang

is an undergraduate student of Department of Clinical Skills Training Center, Zhujiang Hospital, Southern Medical University. Her research interests include medical education.

Jiacheng Wang

is a Master student of Faculty of Artificial Intelligence in Education, Central China Normal University. His research interests include learning analytics and educational data mining.

Ye Yu

is a Master student of Faculty of Artificial Intelligence in Education, Central China Normal University. Her research interests include learning analytics and educational data mining.

Shudi Li

is a Master student of Faculty of Artificial Intelligence in Education, Central China Normal University. His research interests include learning analytics and educational data mining.

Bowen Liu

is an associate professor at Faculty of Artificial Intelligence in Education in Central China Normal University. His research interests include metacognition, self-directed learning and learning analytics.

Jing Tian

is the director of Department of Clinical Skills Training Center, Zhujiang Hospital, Southern Medical University. His research interests include medical education, educational neuroscience and orthopedics.

Authors’ contributions

LH contributed to the research idea and study design. LH and YYZ conducted data collection and wrote the manuscript. JCW、YY and SDL conducted statistical analyses. JCW、YY and SDL edited and revised the paper. WL、JT and BWL led the supervision of the project. All authors read and approved the final manuscript.

Funding

The authors reported there is no funding associated with the work featured in this article.

Data availability

The data that support the findings of this study are available from the corresponding author, [JT], upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by Central China Normal University, Ethic Committee, EC, Institutional Review Board(CCNU-IRB-202409002 A). We also obtained informed consent from participants for the study through an online questionnaire. I confirm that all methods were performed in accordance with the relevant guidelines. All procedures were performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.