Content area
Aim
This study compares the impact of virtual reality (VR) training versus conventional e-learning on newly graduated nurses (NGNs) learning six core clinical skills. We primarily assessed self-efficacy, along with secondary outcomes including knowledge, skills and satisfaction, while also identifying factors that influence self-efficacy.
Background
NGNs need strong self-efficacy and skills for clinical transitions. VR’s immersive training potential is promising, but its long-term impact versus traditional methods is unclear due to mixed evidence.
Design
Quasi-experimental pre-test/post-test with a comparison group and three-month follow-up.
Methods
150 NGNs at a Taiwanese medical center were assigned to VR (n = 75) or e-learning (n = 75) groups via alternating cohorts. Both received physical skills training post-online modules. Outcomes were assessed at baseline (T1), post-physical training (T2) and three months (T3) using the General Self-Efficacy Scale, cognitive questionnaire, Direct Observation of Procedural Skills and satisfaction scale. Mixed-design ANOVA and regression analyzed data, with last observation carried forward for attrition.
Results
No significant Time*Group interactions were found for self-efficacy (p = 0.970), cognitive knowledge (p = 0.459), clinical skills (p = 0.741), or satisfaction (p = 0.245), showing VR was not superior. Significant Time effects (p < 0.001) indicated T1-to-T2 gains, with declines at T3 for self-efficacy and cognition. Baseline self-efficacy and T3 satisfaction predicted T3 self-efficacy (R²=0.215). Attrition was 42.7 % by T3.
Conclusions
VR training did not outperform e-learning over three months. Both supported short-term gains, but sustaining these requires further strategies. Baseline self-efficacy and satisfaction are key predictors. Future studies should address attrition and optimize VR designs for NGNs.
1 Introduction
Virtual reality (VR) is increasingly recognised as a transformative tool in nursing education, providing immersive and risk-free environments for the practice of essential clinical skills without jeopardising patient safety ( Vogelsang et al., 2024). For newly graduated nurses (NGNs), who often encounter significant challenges when transitioning into complex clinical settings due to limited hands-on experience and workplace stress ( Almaqbali and Alnassri, 2024), innovative and effective training methodologies are paramount. Such methods are critical for equipping NGNs with the requisite competence and confidence to deliver safe, high-quality patient care ( Charette et al., 2023). Central to this preparation is the development of self-efficacy—an individual’s belief in their capacity to execute tasks successfully ( Bandura, 1986)—which serves as a cornerstone of effective nursing education. Enhanced self-efficacy empowers NGNs to better navigate clinical challenges, persevere in skill acquisition and adapt to the dynamic healthcare landscape ( Alruwaili et al., 2025).
VR training, characterised by interactive simulations and immediate feedback, offers a promising avenue for cultivating self-efficacy by enabling repeated practice in controlled scenarios ( Ismailoglu and Zaybak, 2018). Theoretically, VR's potential to enhance self-efficacy more effectively than traditional methods can be explained through Bandura's (1986) four principal sources of self-efficacy. VR provides immersive opportunities for "mastery experiences" by allowing learners to repeatedly practice skills in a safe, consequence-free environment until proficiency is achieved. The interactive, first-person perspective offers a more potent form of learning than the passive "vicarious experience" of watching an e-learning video. Furthermore, the immediate, corrective feedback from the system can act as a form of "verbal persuasion," while the controlled, simulated setting may reduce the performance anxiety ("affective states") often associated with initial hands-on practice, thereby fostering a more positive learning experience.
Indeed, emerging evidence suggests that VR can augment knowledge, technical proficiency and learner satisfaction beyond conventional training approaches ( Lin et al., 2024; Padilha et al., 2019). For example, studies have shown improvements in nursing students' self-efficacy for intravenous catheter insertion and enhanced decision-making accuracy in neonatal intensive care nurses through VR ( Ismailoglu and Zaybak, 2018; Alruwaili et al., 2025). Notwithstanding these promising developments, the literature presents inconsistencies regarding the precise impact of VR on nursing self-efficacy ( Lin et al., 2024; Liu et al., 2023). While some investigations report significant gains in self-efficacy ( Ismailoglu and Zaybak, 2018; Lin and Chen, 2021), others find limited or negligible improvements when compared with traditional educational modalities ( Padilha et al., 2019; Lin et al., 2024). A recent systematic review, for instance, indicated that while VR significantly boosted confidence, its effect on self-efficacy was less pronounced, potentially influenced by variables such as VR system design, individual learner attributes, or the duration of assessment ( Lin et al., 2024). Furthermore, the sustained impact of VR on self-efficacy, especially for NGNs during their critical transition to practice, remains an under-investigated area ( Vogelsang et al., 2024). Technical challenges, including suboptimal visual feedback or the potential for motion sickness, can also impede the learning experience ( Hur et al., 2025). Crucially, a paucity of research examining the longitudinal effects of VR training restricts a comprehensive understanding of its enduring benefits—a significant lacuna, considering the evolving nature of NGN confidence with increasing clinical exposure ( Almaqbali and Alnassri, 2024).
Addressing the identified gaps in the literature, the primary aim of this study was to evaluate the effects of a VR-based training program, compared with conventional e-learning, on the primary outcome of self-efficacy among NGNs learning six core clinical skills (blood transfusion, urinary catheterization, intravenous cannulation, suctioning, oral medication administration and intravenous pump use) within a Taiwanese medical center. The secondary objectives were: (1) to compare the effects of the two training modalities on the secondary outcomes of cognitive knowledge, clinical skills and satisfaction; (2) to examine the sustained impact of both interventions over a three-month follow-up period; and (3) to identify factors predicting NGNs' self-efficacy.
Based on the purported benefits of immersive learning, we hypothesized that NGNs in the VR training group would demonstrate significantly greater improvements in self-efficacy, cognitive knowledge, clinical skills and satisfaction compared with those in the conventional e-learning group and that these effects would be sustained at the three-month follow-up. By pursuing these objectives, this research seeks to contribute evidence that can inform the development of nursing education curricula internationally, ultimately aiming to better prepare NGNs for the complexities of contemporary clinical practice.
2 Methods
2.1 Study design
This study used a quasi-experimental pre-test/post-test design with a comparison group. NGNs were allocated to either a VR training group (experimental) or a conventional e-learning group (control) using an alternating cohort sequence. Due to the distinct nature of the interventions (VR headsets vs. e-learning videos), blinding of participants, educators and outcome assessors was not feasible.
Outcome data were collected at three time points: baseline (T1, July 2023), immediately after completion of the physical-skills workshops (T2, July – August 2023) and at a three-month follow-up (T3, October – November 2023). The T2 data collection was intentionally timed to capture the outcomes following the physical workshop, as the study aimed to evaluate which online module (VR or e-learning) better prepared NGNs for this mandatory, hands-on component of their standard hospital training. A planned six-month follow-up was cancelled due to high attrition. This study adheres to the TREND statement for quasi-experimental studies ( Supplementary Table 1) ( Des Jarlais et al., 2004).
2.2 Participants
The study was conducted at a Taiwanese medical center, recruiting NGNs employed between May and July 2023. Inclusion criteria were: (1) registered nurses in their first post-graduate year (PGY1); (2) employed full-time in units requiring proficiency in six core clinical skills (blood transfusion, urinary catheterization, intravenous cannulation, suctioning, oral medication administration and intravenous pump use); and (3) no prior formal VR training experience. Exclusion criteria included: (1) nurses not in PGY1; (2) nurses transferred from other facilities with prior experience; and (3) nurses in specialized units where the targeted skills were not routinely performed.
Using G*Power 3.1.9.4, a sample size of 120 participants (60 per group) was calculated based on a medium effect size (d=0.5), α= 0.05 and 80 % power, accounting for 10 % attrition. To accommodate potential dropout, 150 NGNs were recruited (75 per group) at T1. Recruitment occurred during the hospital’s orientation program for new hires. After providing informed consent, eligible NGNs were randomized to the experimental or control group using an alternating allocation method across training cohorts. Between the T2 and T3 assessments, all participants were actively engaged in clinical practice in their assigned units, where they were required to perform the six targeted procedural skills as part of their standard nursing responsibilities.
2.3 Interventions
Both groups received training on six core clinical skills: blood transfusion, urinary catheterization, intravenous cannulation, suctioning, oral medication administration and intravenous pump use. The training for all participants consisted of two stages: an initial online module followed by an identical, standardized physical skills workshop conducted at the clinical skills center. All NGNs completed their assigned online module before attending the physical skills training. The key experimental difference between the groups lay in the modality of the online training component.
2.3.1 Online training component
Participants in both groups were granted unlimited access to their respective online training materials from enrollment until the T3 follow-up. They were required to complete all assigned modules before attending the physical skills workshop but were encouraged to repeat the modules as needed to reinforce their learning. To ensure content parity, the core educational material for both the VR modules and the e-learning videos was developed based on the same standardized clinical protocols and procedural guidelines established by the medical center.
2.3.1.1 Experimental group (VR training)
Participants used VR modules developed by HTC in 2022, delivered via the Virti platform using VIVE Focus 3 head-mounted displays (
2.3.1.2 Control Group (Conventional online training)
Participants accessed six 30-minute e-learning videos per skill via the hospital’s digital platform, followed by an online knowledge test for each skill.
2.3.2 Physical skills training component
Following the online modules, all participants attended identical hands-on workshops at the clinical skills center, delivered by trained clinical educators to ensure consistency. These sessions integrated the Direct Observation of Procedural Skills (DOPS) assessments for T2. A flipped classroom approach was used for five skills, while a traditional teaching approach (demonstration and return demonstration) was used for the blood transfusion skill.
2.4 Measures
Participant outcomes were assessed using four primary instruments:
• Cognition: Knowledge pertaining to clinical protocols for the six targeted skills was evaluated using a 30-item true/false questionnaire (5 items per skill). This instrument was developed by the research team (the full questionnaire is provided in Supplementary Table 2) and piloted with 15 NGNs not involved in the main study, demonstrating acceptable internal consistency (Cronbach’s α = 0.85). Total scores ranged from 0 to 30, with higher scores indicating greater knowledge.
• Clinical Skills (Psychomotor): Procedural competence was assessed using the DOPS method. Skill-specific checklists, adapted from established hospital standards, were used for all six skills. The maximum attainable scores for each skill were: suctioning (115), oral medication administration (130), intravenous pump use (165), urinary catheterization (110), blood transfusion (150) and intravenous cannulation (160), resulting in a cumulative maximum possible score of 790. DOPS assessments were scored by trained clinical instructors who had achieved satisfactory inter-rater reliability (Intraclass Correlation Coefficient [ICC] = 0.88).
• Self-Efficacy (Affective): The Chinese version of the General Self-Efficacy Scale (GSES; Zhang and Schwarzer, 1995) was administered. This 10-item scale employs a 4-point Likert-type response format (ranging from 1 =not at all true to 4 =completely true). The GSES has established reliability and validity and in this study, it demonstrated high internal consistency (Cronbach’s α = 0.90). Total scores range from 10 to 40, with higher scores reflecting greater self-efficacy.
• Satisfaction (Affective): Overall satisfaction with the training program was measured using a single-item, 10-point Likert scale (ranging from 1 =least satisfied to 10 =most satisfied). The use of such single-item scales for assessing satisfaction in educational and clinical placement settings has been supported in the literature ( Vivanti, Haron, and Barnes, 2014). Satisfaction was assessed at T2 and T3 only.
2.5 Data collection
Data were collected at three time points:
T1 (July 2023): Participants completed online questionnaires for cognition and self-efficacy on their first day.
T2 (July–August 2023): Post-physical skills training, participants completed questionnaires for cognition, self-efficacy and satisfaction. DOPS assessments for all six skills were conducted by skills center instructors during training.
T3 (October–November 2023): In clinical units, participants completed DOPS assessments guided by unit clinical educators, alongside questionnaires for cognition, self-efficacy and satisfaction. Where possible, the same instructor evaluated a participant at T2 and T3. Data were stored in a password-protected database, anonymized before analysis.
2.6 Statistical analysis
Data were analyzed using SPSS 29.0. Descriptive statistics summarized participant characteristics and outcomes. The Shapiro-Wilk test assessed data normality. Baseline comparability (T1) was examined using independent t-tests for continuous variables and chi-square tests for categorical variables. A mixed-design ANOVA with Bonferroni correction compared group differences (experimental vs. control) in cognition, self-efficacy and satisfaction across T1–T3 (T2–T3 for satisfaction) and DOPS scores across T2–T3, reporting effect sizes (partial η²). Repeated-measures ANOVA assessed within-group changes. Multiple regression explored factors influencing self-efficacy, controlling for baseline scores and clinical experience. Missing data at T3 were handled using last observation carried forward (LOCF), with sensitivity analyses comparing LOCF to complete-case results. Significance was set at p < 0.05, with 95 % confidence intervals reported.
2.7 Ethical Considerations
The study was approved by the medical center’s Institutional Review Board (IRB No. 220319). Participants provided written informed consent, informed of the study’s purpose, procedures and voluntary nature. Withdrawal was permitted without consequences. Data were anonymized and stored securely, adhering to the Declaration of Helsinki.
3 Results
3.1 Participant characteristics
Of 200 screened NGNs, 150 were enrolled and allocated to the VR group (n = 75) or the e-learning control group (n = 75), as detailed in the participant flow diagram (
3.2 Primary outcome – self-efficacy
Analysis using mixed-design ANOVA revealed a significant main effect of time on GSES scores (listwise: p < 0.001, partial η² = 0.231; LOCF: p < 0.001, partial η² = 0.104), indicating that scores for all participants, regardless of group, changed significantly across the three time points. Mean scores increased from baseline (T1) to post-training (T2) but then declined at the three-month follow-up (T3). Crucially, there was no significant time*group interaction effect (listwise: p = 0.876; LOCF: p = 0.970), which demonstrates that the pattern of self-efficacy change over time was not different between the VR and control groups (
Multiple linear regression identified baseline GSES scores and T3 satisfaction as the only significant positive predictors of T3 self-efficacy in both listwise (Adjusted R² = 0.136) and LOCF analyses (Adjusted R² = 0.182) ( Supplementary Table 3).
3.3 Secondary outcomes – cognition, clinical skills and satisfaction
LOCF analyses for all secondary outcomes showed patterns consistent with the primary outcome: significant time effects but no significant time*group interaction effects ( Table 2).
Cognitive Knowledge: Scores improved significantly from T1 to T2 and remained above baseline at T3 (p < 0.001).
Clinical Skills (DOPS): DOPS scores continued to significantly increase from T2 to T3 (p < 0.001), reflecting skill improvement during clinical practice.
Training Satisfaction: Satisfaction was high at T2 but decreased modestly by T3 (p < 0.001).
3.4 Data integrity
The study experienced a 42.7 % attrition rate by T3. Sensitivity analyses using LOCF imputation produced findings concordant with complete-case analyses, supporting the robustness of the study's conclusions despite the high attrition. The planned six-month follow-up was cancelled, precluding inferences about longer-term outcomes.
4 Discussion
This quasi-experimental study found that a VR-based training program did not produce superior outcomes in self-efficacy, cognitive knowledge, clinical skills, or satisfaction compared with conventional e-learning among NGNs over a three-month period. The lack of significant Time*Group interactions across all outcomes suggests that both preparatory modalities resulted in similar learning trajectories. While both groups showed short-term gains post-training in self-efficacy and cognition, these were often not sustained at the three-month follow-up ( Vogelsang et al., 2024). In contrast, clinical skills (DOPS) uniquely continued to improve from T2 to T3, likely reflecting skill consolidation through real-world clinical practice. Furthermore, while the training modality itself did not predict outcomes, our regression analysis revealed that baseline self-efficacy and training satisfaction were significant predictors of NGNs' subsequent self-efficacy, highlighting the importance of pre-existing attributes and learning experiences ( Bandura, 1986). These findings contribute to a more nuanced understanding of VR's role in nursing education ( Liu et al., 2023).
The absence of a VR advantage in self-efficacy aligns with systematic reviews noting that VR's impact on this stable belief is often less pronounced than its effect on transient confidence ( Lin et al., 2024). Our study suggests that even immersive VR did not cultivate a stronger belief in task performance than e-learning when both were followed by an identical, powerful learning stimulus: standardized physical skills training. The pattern of short-term gains followed by a decline underscores the challenge of retention and the need for ongoing support beyond initial training.
Several factors may explain the lack of VR superiority. The hands-on workshop may have acted as a powerful equalizer, overriding any unique benefits of the preceding online module. Technical issues, including two withdrawals due to dizziness from VR, could have also diminished the learning experience ( Hur et al., 2025). Finally, the quasi-experimental design and high attrition rate (42.7 %), though handled with sensitivity analyses, are significant methodological considerations.
4.1 Implications for nursing education
Our findings suggest that the mere adoption of advanced technology like VR is not a panacea; the underlying instructional design and, crucially, the opportunity for physical practice remain paramount ( Chen et al., 2020). Before making substantial financial investments in VR, institutions should consider that high-quality, conventional e-learning remains a cost-effective and powerful tool for delivering foundational knowledge ( Shorey and Ng, 2021). The key takeaway is not that VR is ineffective, but that its value may require more sophisticated instructional design or lie in specific applications not captured here. Furthermore, the decline in outcomes by T3 highlights an urgent need for structured post-orientation support. Incorporating strategies like regular "booster sessions," mentorship and consistent clinical feedback could be essential for consolidating and maintaining NGNs' skills and self-efficacy long-term ( Almaqbali and Alnassri, 2024).
4.2 Strengths of the study
Despite the null findings, this study has several strengths. Its longitudinal design with a three-month follow-up provides a rare perspective on the sustainability of training effects. The study was conducted in a pragmatic, real-world setting with NGNs, enhancing its ecological validity. We also employed a comprehensive set of outcome measures across affective, cognitive and psychomotor domains. Finally, the publication of these rigorous null findings is itself a strength, helping to mitigate publication bias and encouraging a more critical, evidence-based discourse on the adoption of expensive educational technologies.
4.3 Limitations and future research directions
Several limitations inherent in this study warrant consideration and concurrently illuminate avenues for future investigation. Notably, the substantial attrition rate (42.7 %) at the three-month follow-up, primarily due to NGNs' workload and scheduling conflicts, reduced the sample size for listwise analyses. Despite analytical mitigation through LOCF, this may have introduced bias and diminished statistical power to detect subtle group differences. This underscores the need for future studies to incorporate larger, multi-center samples, which would not only enhance statistical power and the generalizability of findings beyond the current single-center Taiwanese context but also necessitate robust strategies to maximize participant retention. The logistical constraints preventing a planned six-month follow-up also restricted insights into the longer-term sustainability of training effects, a critical gap highlighted by Vogelsang et al. (2024), thus emphasizing the importance of future longitudinal assessments with more extended follow-up periods, such as six, twelve, or even twenty-four months.
Furthermore, while the quasi-experimental design with alternating cohort allocation was a pragmatic choice, it carries an inherent risk of selection bias not present in true randomized controlled trials (RCTs); future research would benefit from employing robust RCT designs with individual participant randomization ( Alruwaili et al., 2025). Additionally, the frequency and context of skills practice for each NGN in their clinical units between T2 and T3 were not monitored, representing a potential unmeasured confounding variable that could influence skill consolidation and self-efficacy. The standardized physical skills training component, common to both intervention and control groups, likely provided a potent learning experience that may have masked or diluted any unique contributions of the preceding VR module. Subsequent research could explore VR's impact with different post-online training protocols or compare various VR instructional designs ( Huai et al., 2024). To this end, investigations into VR design improvements, such as adaptive feedback mechanisms, gamification elements, or scenarios tailored to specific NGN challenges, may enhance effectiveness ( Hur et al., 2025).
Moreover, to better understand the inconsistent effects of VR noted in the literature, future studies should delve into how learner characteristics (e.g., technological proficiency, learning styles) and contextual factors (e.g., workplace support systems) interact with VR training outcomes ( Obeid et al., 2025). Complementing quantitative data with qualitative methodologies will also be crucial for gaining deeper insights into NGNs' subjective experiences with VR, including aspects of engagement, perceived realism and overall satisfaction, which can inform the development of more learner-centric interventions. Finally, conducting comprehensive cost-effectiveness analyses will be essential to guide institutional decisions regarding the adoption and scalability of VR technologies in nursing education.
5 Conclusion
This study compared the effects of a VR-based training program with conventional online training on the self-efficacy, cognitive knowledge, clinical skills and training satisfaction of NGNs in a Taiwanese medical center. Over a three-month period, the VR intervention did not demonstrate statistically significant advantages over the conventional approach for any of the measured outcomes. While both training modalities facilitated short-term improvements in most areas immediately post-intervention, these gains were not always sustained at the three-month follow-up, highlighting a need for ongoing learning reinforcement. Baseline self-efficacy and satisfaction with training were identified as important predictors of NGNs' subsequent self-efficacy.
The findings contribute to a nuanced understanding of VR's application in nursing education, suggesting that its current iterations may not inherently surpass well-structured conventional training, especially when both are followed by practical skills reinforcement. The challenges of high participant attrition and the inability to conduct a longer-term follow-up temper the conclusions that can be drawn about sustained effects. Nursing educators and institutions should therefore adopt a discerning approach to integrating VR, focusing on optimizing its design, ensuring it aligns with specific learning objectives and supplementing it with strategies to foster long-term retention and application of learning. Future research, incorporating rigorous designs, larger samples and extended follow-up periods, is essential to fully unpack the potential of VR in preparing NGNs for the complexities of contemporary clinical practice.
CRediT authorship contribution statement
WanRu Huang: Validation, Resources, Project administration, Investigation, Formal analysis, Data curation. ShuJyuan Chen: Resources, Project administration, Investigation, Data curation. YaWen Lee: Writing – review & editing, Writing – original draft, Supervision, Conceptualization. Chihhao Lin: Writing – original draft, Validation, Software, Formal analysis. AiLing Chang: Resources, Investigation. HuiHsin Ku: Validation, Resources, Project administration, Investigation, Formal analysis, Data curation.
Funding
This research was supported by the
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We would like to thank all the clinical instructors, physicians, nurses, and relevant staff who participated in this study.
Appendix A Supporting information
Supplementary data associated with this article can be found in the online version at
Appendix A Supplementary material
Supplementary material
Supplementary material
Supplementary material
Table 1
| Characteristic | VR Group (n = 75) | Control Group (n = 75) | p-value |
| Age (years, Mean ± SD) | 24.3 ± 3.4 | 24.5 ± 2.8 | 0.81 |
| Gender, n (%) | 0.13 | ||
| Female | 66(88.0) | 58(77.3) | |
| Male | 9(12.0) | 17(22.7) | |
| Education, n (%) | 0.81 | ||
| Bachelor’s Degree | 64(85.3) | 66(88.0) | |
| Associate Degree | 11(14.7) | 9(12.0) | |
| Unit Assignment, n (%) | 0.48 | ||
| General Ward | 54(72.0) | 49(65.3) | |
| Intensive Care Unit | 21(28.0) | 26(34.7) |
Table 2
| Outcome | Group | T1
(M±SD) |
T2
(M±SD) |
T3
(M±SD) |
Time Effect
(p, η²) |
Time*Group Interaction (p, η²) |
| Self-Efficacy (Primary) | VR | 25.4 ± 4.8 | 27.3 ± 6.0 | 24.5 ± 5.5 | < 0.001, 0.104 | 0.970, 0.001 |
| Control | 25.0 ± 4.3 | 26.9 ± 6.5 | 24.3 ± 6.5 | |||
| Cognition (Secondary) | VR | 14.9 ± 2.6 | 18.9 ± 3.5 | 17.8 ± 3.5 | < 0.001, 0.426 | 0.459, 0.005 |
| Control | 14.7 ± 2.0 | 18.3 ± 2.7 | 17.8 ± 2.8 | |||
| Clinical Skills (DOPS) | VR | - | 639 ± 21.0 | 652.9 ± 32.8 | < 0.001, 0.093 | 0.741, 0.001 |
| Control | - | 645.6 ± 15.7 | 657.3 ± 43.8 | |||
| Satisfaction | VR | - | 8.8 ± 1.1 | 8.4 ± 1.2 | < 0.001, 0.083 | 0.245, 0.009 |
| Control | - | 9.0 ± 1.1 | 8.8 ± 1.2 |
© 2025 The Authors