Virtual reality-based training improves thoracentesis skills in medical interns: a randomized controlled trial

Abstract

Background

Thoracentesis is an essential clinical procedure, but its teaching is often limited by patient safety concerns and insufficient opportunities for repeated practice. Virtual reality (VR) offers immersive, repeatable simulation-based training; however, its effectiveness for thoracentesis has not been rigorously evaluated.

Methods

In this randomized controlled trial, 20 medical interns were randomly assigned to either a VR-based training group (n = 10) or a traditional training control group (n = 10). The VR group received theoretical instruction plus VR simulation practice (1 h/day, 4 days/week for 3 weeks), while the control group received the same theoretical instruction plus traditional mannequin-based practice. All participants completed a 4-week training program. Outcomes were assessed at baseline, week 2, week 3, and week 4 using a standardized 300-point scoring rubric (preoperative, intraoperative, and postoperative components). Statistical comparisons were made using t-tests and chi-square tests. The primary outcome was the final thoracentesis procedural score at week 4.

Results

No adverse events occurred. Baseline characteristics and initial assessment scores did not differ significantly between groups. At week 2, the VR group scored significantly lower than the control group (p < 0.05), reflecting an initial learning curve. However, at week 3 and week 4, the VR group significantly outperformed the control group (p < 0.01). At the final assessment, 80% (8/10) of the VR group achieved scores ≥ 270 points (excellent), compared to only 10% (1/10) of the control group.

Conclusion

VR-based training may improve thoracentesis procedural skills after an initial adaptation period. VR appears to be a useful adjunct to traditional medical education, though these findings are preliminary due to the small sample size.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12909-026-09801-8.

Background

Thoracentesis (pleural puncture) is a fundamental invasive procedure used to diagnose and manage pleural effusion and pneumothorax. It involves inserting a needle into the pleural space to aspirate fluid or air. Despite being a common procedure, thoracentesis carries risks, including pneumothorax, hemothorax, infection, and injury to intercostal nerves or blood vessels. Therefore, competent technical skill is essential for patient safety.

Traditionally, thoracentesis is taught through lectures, demonstrations, and supervised practice on patients or mannequins. However, several barriers limit effective teaching: (1) patient safety concerns restrict student practice opportunities; (2) increasing concerns regarding patient safety and procedural consent reduce bedside teaching opportunities; (3) complications cannot be deliberately practiced; and (4) traditional mannequin-based training has inherent limitations [1, 2]. Consequently, many medical graduates report insufficient confidence and competence in performing thoracentesis independently [3].

Virtual reality (VR) technology creates immersive, interactive, three-dimensional environments that allow learners to practice procedures repeatedly without patient risk. VR has been increasingly applied in various medical education contexts [4, 5]. Studies suggest that VR training can improve hand-eye coordination, procedural memory, and clinical reasoning [6]. However, its application in thoracentesis training remains limited and lacks rigorous evaluation. Recent systematic reviews have confirmed the effectiveness of VR in improving procedural skills across various medical disciplines, though evidence specific to thoracentesis remains limited [5, 6]. VR is well-suited to support deliberate practice—a key component of skill acquisition—by allowing learners to repeat procedures with real-time feedback in a low-stress environment [7].

This study aimed to evaluate the effectiveness of VR-based training compared to traditional mannequin-based training in improving thoracentesis skills among medical interns. We hypothesized that VR training would lead to superior procedural competence after a 4-week training period.

Methods

Study design and participants

This was a single-center, parallel-group, randomized controlled trial conducted between March 2023 and March 2024. Twenty medical interns rotating through the Department of Respiratory Medicine at Huai’an Hospital Affiliated to Yangzhou University were enrolled. In the Chinese medical education system, students typically enter a 3-year junior college or 4-year undergraduate medical program after high school. The interns in this study were in their final year of training, with an average age of 19 years. All eligible interns (n = 22) were invited to participate; two declined due to scheduling conflicts. A CONSORT flow diagram is provided as Fig. 1.

Fig. 1.

A CONSORT flow diagram

Inclusion criteria

(1) internship duration ≥ 4 weeks; (2) completion of basic clinical coursework; (3) informed consent provided.

Exclusion criteria

(1) refusal to participate; (2) internship duration < 4 weeks; (3) early withdrawal from the internship.

Randomization and blinding

Participants were randomly assigned (1:1) to either the VR training group or the control group using a computer-generated random number table. The randomization sequence was generated by a research assistant not involved in participant training or assessment. Allocation was concealed using sequentially numbered, opaque, sealed envelopes. The instructor who delivered theoretical lectures and supervised practice was not blinded due to the nature of the intervention. However, the assessor who conducted all final skill evaluations was blinded to group allocation.

VR system development

A custom VR thoracentesis simulator was developed for this study. Chest CT and ultrasound images from anonymized patients were imported into Yingtong 3D software (Yingtong Technology, China) to reconstruct a three-dimensional chest model. The model was then imported into Unity software (Unity Technologies, USA) for rendering, interactive function development (e.g., needle insertion angle, depth, aspiration), and real-time feedback. The final VR module was deployed on the Yingtong VR platform (HTC Vive headset and hand controllers). The system provided visual guidance, anatomical labeling, and performance feedback (e.g., correct vs. incorrect needle placement). The VR system provided visual feedback on needle insertion angle (target: 90° to the skin), depth (target: 3–5 cm), and aspiration success. However, the system did not provide haptic (force) feedback, which 20% of participants later identified as a limitation.

Training protocol

All participants completed a 4-week training program. Both groups attended a 1-hour standardized theoretical lecture covering thoracentesis indications, contraindications, anatomy, procedural steps, and complications. Baseline assessment occurred at week 1 before any practical training. Practical training occurred during weeks 2, 3, and 4.

From week 2 to week 4, the two groups received different practical training:

VR group

Each participant practiced independently on the VR simulator for 1 h per day, 4 days per week (Monday to Thursday, 16:00–17:00). The simulator provided real-time feedback on performance. Participants could repeat procedures unlimited times.

Control group

Participants practiced on a traditional thoracentesis mannequin in groups of five under the supervision of the same instructor, for 1 h per day, 4 days per week. The instructor demonstrated, observed, and corrected errors.

Outcome measures

Assessments were conducted at four time points: baseline (before practical training, week 1), mid-point (week 2), week 3, and final (week 4). The assessment used a standardized 300-point scoring rubric divided into three domains:

Preoperative preparation (100 points)

Patient identification, positioning, informed consent, sterile technique, local anesthesia.

Intraoperative operation (100 points)

Needle insertion site selection, angle, depth control, aspiration technique.

Postoperative operation (100 points)

Specimen handling, wound care, complication recognition, documentation.

All assessments were performed by the same senior instructor, who was blinded to group allocation. The scoring rubric was pilot-tested on five senior residents (not part of the study) by two independent raters to assess inter-rater reliability (kappa = 0.89). In the main study, all assessments were performed by a single senior instructor who was blinded to group allocation.

Additionally, a post-training survey was developed by the research team to evaluate participants’ perceptions of VR teaching. The survey consisted of six statements rated on a 5-point Likert scale (1 = strongly disagree to 5 = strongly agree), covering learning interest, autonomous learning ability, innovative thinking, interactivity, course content richness, and tactile feedback realism. The survey questionnaire is provided in Supplementary File 1.

The scoring rubric was developed by three respiratory medicine faculty members based on institutional thoracentesis competency checklists. Content validity was assessed by two external experts (senior pulmonologists not involved in the study), who rated each item as “essential” or “useful” (agreement: 92%). The post-training survey was developed by the research team for this study and was not formally validated. Results should be considered exploratory.

Statistical analysis

Based on a pilot study, we estimated that a sample size of 10 per group would provide 80% power to detect a large mean difference of 30 points (SD 20, Cohen’s d = 1.5) with a two-sided alpha of 0.05. The study was not powered to detect small or moderate effect sizes, which is a limitation.

Continuous variables were expressed as mean ± standard deviation (SD) and compared using independent samples t-tests (normality confirmed by Shapiro-Wilk test). Categorical variables were compared using chi-square or Fisher’s exact test as appropriate. Repeated measures were analyzed using two-way ANOVA with post-hoc Bonferroni correction. A p-value < 0.05 was considered statistically significant. All analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA) and GraphPad Prism version 9.5 (GraphPad Software, San Diego, CA, USA). Effect sizes (Cohen’s d) were calculated for between-group comparisons at week 4 and are reported in the Results section.

Ethics approval

This study was approved by the Institutional Review Board of Huai'an Hospital Affiliated to Yangzhou University (approval number: KY-P-2023-083-01). All participants provided written informed consent. The study was conducted in accordance with the Declaration of Helsinki

Results

Participant characteristics

Twenty participants completed the study (no withdrawals). Baseline characteristics were well-balanced between the two groups (Table 1). No significant differences were found in age, sex, internship duration, or educational level (p > 0.05 for all).

Table 1.

Baseline characteristics of participants

Characteristic	Control group (n = 10)	VR group (n = 10)	Test value	p-value
Sex (Male/Female)	6 / 4	5 / 5	χ² = 0.202	0.653
Age (years, mean ± SD)	19.10 ± 1.10	19.40 ± 0.70	t = 0.728	0.476
Internship weeks (mean ± SD)	7.70 ± 1.70	8.20 ± 1.32	t = 0.735	0.472
Education (Junior college/Undergraduate)	6 / 4	6 / 4	χ² = 0.175	0.675

Skill assessment scores

All participants completed the training without adverse events. Assessment scores across the four time points are shown in Fig. 2.

Fig. 2.

Comparison of thoracentesis scores between VR and control groups over the 4-week training period

Baseline (week 1)

No significant difference between groups (VR: 185.4 ± 12.3, control: 188.2 ± 11.6; p = 0.612).

Week 2

The VR group scored significantly lower than the control group (VR: 205.3 ± 14.2, control: 225.7 ± 10.5; p = 0.003), indicating an initial learning curve with the VR system.

Week 3

The VR group caught up and showed no significant difference (VR: 248.6 ± 11.4, control: 240.3 ± 12.1; p = 0.131).

Week 4 (final)

The VR group significantly outperformed the control group (VR: 278.4 ± 9.8, control: 252.1 ± 10.3; p < 0.001). The between-group difference at week 4 corresponded to a large effect size (Cohen’s d = 2.60, 95% CI: 1.35 to 3.85).

Data are presented as mean ± SD. ***p < 0.001, **p < 0.01 compared to control group at the same time point (two-way ANOVA with Bonferroni correction).

At the final assessment, 8 participants (80%) in the VR group achieved an “excellent” score (≥ 270 points), compared to only 1 participant (10%) in the control group (p = 0.005, Fisher’s exact test). Conversely, 9 participants (90%) in the control group scored in the “good” range (240–269 points), compared to 2 (20%) in the VR group. The cutoff of ≥ 270 points for “excellent” was defined post-hoc based on the distribution of final scores, as no published standard exists.

Participant perceptions of VR teaching

Post-training survey results (Table 2) showed high acceptance of VR technology among the VR group. Notably, 90% agreed or strongly agreed that VR improved autonomous learning ability, and 80% reported enhanced learning interest and interactivity. However, 20% disagreed that VR provided realistic tactile feedback, suggesting a need for supplementary hands-on practice. The survey was administered by a research assistant not involved in training, but social desirability bias cannot be ruled out, as participants may have wished to please the research team.

Table 2.

Student perceptions of VR teaching (VR group, n = 10)

Statement	Agree/Strongly agree (%)
VR stimulates learning interest	80%
VR improves autonomous learning ability	90%
VR enhances innovative thinking	80%
VR system has good interactivity	80%
VR course content is rich	80%
VR lacks realistic tactile feedback (reverse-coded)	20%

Mean Likert scores (1–5) for the six items ranged from 3.8 to 4.5 (SD 0.7–1.0), with tactile feedback receiving the lowest mean score (3.2 ± 1.2)

Discussion

This small randomized trial provides preliminary evidence that VR-based training may improve thoracentesis skills among medical interns, compared to traditional mannequin-based training. After a 4-week training period, 80% of VR-trained participants achieved excellent competency scores, compared to 10% in the control group. However, these findings require replication in larger, multi-center trials.

Interpretation of findings

The initial lower performance of the VR group at week 2 (p = 0.003) is an important observation. This likely reflects the cognitive load associated with learning a new VR interface, including headset navigation, hand controller manipulation, and adapting to visual-spatial cues [7]. Similar adaptation periods have been reported in VR training for laparoscopy and central line placement [8, 9]. Educators should anticipate this initial dip and provide additional orientation sessions.

By week 3, the VR group had caught up with the control group. By week 4, they significantly outperformed them. This pattern suggests that once participants overcome the initial learning curve, VR training may lead to greater skill gains than traditional methods. Several mechanisms may explain this: (1) unlimited repeatable practice without patient risk; (2) real-time, objective feedback; (3) standardized scenarios that ensure all learners encounter the same challenges; and (4) immersive engagement that enhances procedural memory [4, 10].

It is important to note that the VR group practiced independently with unlimited repetitions, while the control group practiced in groups of five, sharing one mannequin. This difference in training intensity—rather than VR per se—may have contributed to the observed differences. Conversely, this also reflects a real-world advantage of VR: it enables individualized, self-paced practice without resource constraints.

Comparison with previous studies

Our findings align with meta-analyses showing that VR training is superior to no training and non-inferior to traditional training for procedural skills [5, 6]. A 2023 meta-analysis similarly concluded that VR is most effective when combined with hands-on practice, supporting our proposed hybrid model [5]. However, few studies have specifically examined thoracentesis. Previous research has reported that VR improved basic arthroscopy skills in orthopedic residents, and highlighted VR’s potential in medical education [6]. Our study extends these findings to thoracentesis—a common, high-stakes bedside procedure.

Limitations

Several limitations should be acknowledged.

First, the sample size was small (n = 20) and from a single institution, limiting generalizability and statistical power to detect modest effect sizes. The study was powered only for a large effect (d = 1.5).

Second, the training intensity was not equivalent between groups. The VR group practiced independently with unlimited repetitions, while the control group practiced in groups of five on a single mannequin. This is the most significant threat to internal validity. We cannot determine whether the superior performance of the VR group was due to VR technology itself or simply to more individualized, repetitive practice. Conversely, this also highlights a practical advantage of VR: it enables resource-efficient, self-paced learning.

Third, we did not assess long-term skill retention (e.g., at 3 or 6 months).

Fourth, the VR system lacked haptic feedback, which 20% of participants noted as a limitation.

Fifth, we did not measure transfer of skills to real patients—the ultimate validity test.

Sixth, the instructor was not blinded during training, though the outcome assessor was blinded.

Seventh, the post-training survey was exploratory and not formally validated.

Eighth, the “excellent” cutoff (≥ 270) was defined post-hoc and lacks external validation.

Implications for medical education

If confirmed in larger studies, VR could serve as a useful adjunct to traditional thoracentesis teaching. A hybrid model—initial VR training for basic procedural familiarity, followed by supervised hands-on practice—may balance the strengths of both approaches. However, recommendations for institutional investment require additional data on cost-effectiveness and long-term outcomes.

Future directions

Future research should: (1) conduct multi-center trials with larger sample sizes; (2) assess long-term skill retention and transfer to real patients; (3) compare VR with other simulation modalities (e.g., augmented reality, hybrid simulators); (4) evaluate cost-effectiveness; and (5) explore the optimal VR-to-traditional practice ratio.

Conclusions

In this small randomized trial, VR-based training was associated with better thoracentesis procedural skills compared to traditional mannequin-based training after a 4-week program. Despite an initial learning curve, VR enhanced learning efficiency and engagement. A hybrid approach combining VR with limited traditional practice may offer the best educational outcomes, but larger studies are needed to confirm these preliminary findings.

Abbreviations

VR

Virtual Reality

CT

Computed Tomography

SD

Standard Deviation

ANOVA

Analysis of Variance

IRB

Institutional Review Board

Authors' contributions

Conceptualization: SH, ZT, ZC; Methodology: SH, ZT, BC; Software: BC, LW; Validation: ZC, LW; Formal analysis: SH, ZT; Investigation: SH, ZT, BC; Resources: LW, ZC; Data curation: SH, ZT; Writing – original draft: SH, ZT; Writing – review & editing: ZC, BC; Visualization: SH; Supervision: ZC; Project administration: ZC; Funding acquisition: ZC. All authors read and approved the final manuscript.

Funding

This work was supported by Jiangsu Vocational College of Nursing-Integrated Scientific and Educational Clinical Research Project (grant number KRHW2406). The funder had no role in study design, data collection, analysis, interpretation, or manuscript writing.

Data availability

The datasets generated and analyzed during this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Huai’an Hospital Affiliated to Yangzhou University (approval number: KY-P-2023-083-01). Written informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.