Background Virtual reality (VR) applications are effective tools in many educational disciplines. A minimally interactive VR application allowing stereoscopic viewing of surgical videos has been developed to aid veterinary students learning to perform surgery. We sought to describe how students used the VR application while preparing to perform their first canine sterilisation surgery and compare surgical performance of students who prepared using traditional methods with students who also used VR.
Methods Third-year veterinary students (n=44) were randomised into control and VR groups in a parallel superiority randomised controlled trial. All were given lectures, videos and skills practice on models. VR group students were also given a VR application and headset to view stereoscopic surgical videos. Blinded raters scored a subset of students (n=19) as they performed their first canine ovariohysterectomy.
Results and conclusions Groups spent similar time preparing to perform surgery, potentially because of the rigour of students’ non-surgical course load. When VR training was added to an already comprehensive surgical skills curriculum, students watched VR videos for a median of 90 min. Groups did not differ in surgical performance scores or time. A larger study of the VR application with prescribed use guidelines would be a helpful subsequent study.
- preclinical education
- preventive medicine
- soft tissue surgery
- veterinary profession
Statistics from Altmetric.com
Veterinary colleges are required to produce graduates with day one competency in general surgery.1–3 Veterinarians and board-certified surgeons agree that new graduates should possess a wide variety of surgical skills and apply these skills with minimal supervision.4 5 Veterinary education has shifted away from teaching using non-survival surgeries to teaching and assessing skills in a clinical skills laboratory, often on models.6 7 Virtual reality (VR) training offers another non-invasive learning option for surgical skills.
VR immerses users in a computer-generated environment that is viewed three-dimensionally and stereoscopically, where each eye is presented with a slightly different view of the virtual scene, as binocular disparity creates the perception of depth and presence.8 VR allows a variable degree of interactivity using equipment, such as a headset or gloves. Highly interactive and immersive applications can track a user’s movements and map them onto a virtual setting, giving users the experience of walking through the simulation, ‘feeling’ what’s happening with shock sensors, making decisions in the scenario and experiencing the results of their choices. In minimally interactive VR applications, the interaction between users and environment may be limited to a stereoscopic view that is reactive to users’ head movements.
VR applications support learning via constructivism, a theory that posits that the acquisition of knowledge is an active process where individuals construct meaning from their experiences and interactions with their environment.9 10 Although minimally interactive virtual environments do not allow users to learn by doing, they are enhanced by the inclusion of cognitive tools, such as those that support the learner’s ability to visualise a problem in three dimensions, which decreases the cognitive load otherwise required for learners to reconstruct two-dimensional images into a three-dimensional mental rendering.11 12 Virtual environments also allow learners to view objects magnified at close range or at a distance in the context of their surroundings.12
Veterinary educators have called for more effort to be devoted to developing VR training methods.13 Using VR technology to teach surgery supports the three Rs of ethical animal use by potentially replacing systems that previously used non-survival surgeries, reducing the number of animals used to teach surgery and refining students’ skill level before they perform live surgery.14 Many veterinary students preparing to perform their first live surgery report negative emotions, such as feeling nervous, scared or stressed.15 16 These feelings most commonly arise from a lack of self-confidence paired with a strong desire to be well prepared.16 VR offers a potential method of improving student comfort level prior to entering the operating room. At least four veterinary colleges are using minimally interactive VR to teach surgical skills,(http://vetmed.tamu.edu/news/press-releases/first-360-degree-virtual-reality-video-for-veterinary-education-developed and https://www.pyxisvr.com/exero-vet-database-list) but there are no published articles describing the use of VR for teaching veterinary surgery.
We sought to investigate how students preparing to perform their first surgery used and perceived a minimally interactive VR application when it was added to their comprehensive surgical curriculum of lecture, traditional video recordings and repetitive skills practice on a model. Our second objective was to determine whether the students’ use of the VR application improved the quality of their first surgery, specifically a canine ovariohysterectomy. The ovariohysterectomy is a common choice for a student’s first surgery because it requires a celiotomy yet is less complex than other abdominal surgeries commonly performed in general practice.
We hypothesised that students with the VR application would spend more time preparing for their surgery than students who were provided only with traditional methods. We hypothesised that students would report that VR was helpful in their preparation. However, because of the comprehensive nature of the students’ non-VR skills training curriculum, we hypothesised that adding VR would not change the quality of students’ first surgical performance once any difference in time spent in preparation was controlled for.
Enrolment & background
Researchers enrolled a convenience sample of 44 third-year veterinary students preparing to perform their first canine sterilisation surgery as a part of their clinical skills curriculum. Students with a history of motion sickness during VR viewing and those who had previously performed surgery on a live animal were excluded. Enrolled students were randomised into a control group (n=15) and a VR training group (n=29) using a spreadsheet program for this parallel superiority trial. Randomisation into research groups was performed by a research assistant using spreadsheet software; researchers remained unaware of group assignments. Twice as many students were enrolled into the VR group as the control group because we wanted robust feedback evaluating the VR application. These enrolment figures allowed us to survey approximately 25 per cent (29 students) of the entire class of 119 students after their use of the VR application, which we believed would generate ample feedback.
Students participated in the study at the DeBusk Veterinary Teaching Center in Ewing, Virginia in fall 2017. During the preceding five semesters, students had practised surgical skills on task trainers and silicone models, and they had practised the identification of tissue layers, suturing and ligation on unpreserved canine cadaver tissue during clinical skills laboratory sessions (table 1). Laboratory sessions were each 2–3 hours in duration and were supervised at a faculty to student ratio of approximately 1:8. All students had passed objective structured clinical examinations consisting of surgical and non-surgical content each of their first three semesters and had passed a high-stakes surgical skills examination on an ovariohysterectomy model in fourth semester.17
Both groups received non-VR training materials including lectures which could be re-watched online, PowerPoint slides and standard video recordings of sterilisation surgeries hosted on an internet sharing site (YouTube). VR group students were also provided with a VR headset (Voxkin, Kathmandu, Nepal) and complimentary subscription to a minimally interactive veterinary surgical training VR application (Exero Vet, Pyxis, New Orleans, LA) that they downloaded on their smart phones. A student leader was trained to answer questions and solve technological issues related to the application and headsets.
The VR application allowed 330° stereoscopic viewing of digitally recorded veterinary surgeries using students’ smart phones and VR headsets. Students were able to view VR recordings of faculty members at three US veterinary colleges performing surgeries including a live canine castration and a cadaver canine ovariohysterectomy. Students had the option of watching the recordings in three dimensions (stereoscopically) using a VR headset (figure 1) or in two dimensions (monoscopically) using their smart phones (figure 2). In either format, the student could hear an instructional voice-over. In the headset format, the student was able to rotate his or her head and see a corresponding three-dimensional view of items in the surgical suite. Looking directly at the surgical site would toggle the student into a magnified view. When watching in two dimensions without a headset, the student could move his or her phone to alter the view or magnify the surgery but potentially could be more easily distracted by their real-life surroundings. Students received their headsets and application subscription 5–8 weeks before their first surgery. Students were directed to use the VR application in addition to their other resources in the manner they felt was most helpful as they prepared to perform their first live surgery. We chose this over a prescribed use regimen because we wanted to observe how the students would choose to use the application relative to the other resources that we provided to them, without us dictating use requirements.
A veterinary technician not involved with the research assigned students to surgical groups and patients. Students were assigned to surgical teams of three students, where all team members were enrolled in the same study group to prevent the mixing of resources. Each student was randomly assigned to perform the duties of an anaesthetist, primary surgeon or assistant surgeon during three consecutive weeks of live surgeries. Each group of three students was randomly assigned to perform a canine ovariohysterectomy or canine castration for each of the 3 weeks of surgery. Patients were systemically healthy sexually intact mixed-breed canines from a local animal shelter that provided written consent. Veterinary students were assisted in surgery by licensed veterinarians at a ratio of one veterinary instructor per two surgical tables. Instructors offered generalised supervision and guidance, and students were required to check in with instructors at defined surgical checkpoints, including creation of a skin incision, transecting pedicles and completion of body wall closure.
Raters (JAH and SLA) were present in the surgical theatre in addition to instructors. Raters were blinded to students’ group allocation. Raters directly observed a subset of 19 students who performed a canine ovariohysterectomy. Twelve of these students were enrolled in the VR group while seven were in the control group. The number of surgeries observed was limited by the faculty available for rating surgeries, the length of student surgeries and the availability of suitable patients. The raters were experienced surgical skills educators with responsibility for creation and management of surgical skills teaching content and assessments. Raters did not interact with students; all coaching and feedback was provided exclusively by the students’ instructor.
Raters used Read et al’s rubric for canine ovariohysterectomy which scores a student’s performance on 46 surgical steps (online supplementary appendix 1), each scored from 0 (unsatisfactory) to 3 points (excellent) for a possible total of 138 points.18 Each student’s overall surgical skill was also assessed using a global rating scale of 1 (inferior) to 6 (excellent) which took into consideration students’ overall level of surgical skill including efficiency and flow of the procedure, instrument and suture handling, and tissue handling. Raters recorded students’ surgical time to the nearest minute. Patients who experienced haemorrhage that required an instructor to isolate and control were excluded from surgical time analysis.
Students who performed castrations or who performed ovariohysterectomies that were not rated by researchers were included in the post-study surveys about the application and its perceived utility but were excluded from surgical performance evaluation. The 34-question survey given to VR group students asked them to quantify their use of the application and other methods of preparation in the month prior to performing surgery, comment on any issues or challenges they encountered with the VR application or headset, and collected their opinions about the application’s utility and whether they would recommend it to others (online supplementary appendix 2). A few VR group students reported not using the application; they were removed from data analysis. Control group students completed a 19-question survey that asked them to describe and quantify the time they spent preparing for surgery using the various traditional methods (online supplementary appendix 2). Five-point Likert-scale questions were included in both groups’ surveys to compare confidence to perform surgery and self-perceived adequacy of their preparation. Figure 3 is the study workflow.
This study was reviewed and deemed exempt by the Lincoln Memorial University Institutional Review Board. Dogs were provided with a high standard of perioperative veterinary care including preoperative bloodwork; intraoperative anaesthetic monitoring of heart rate, respiratory rate, temperature, blood oxygen saturation, blood pressure and end-tidal carbon dioxide; intravenous fluid administration; and multimodal analgesia.
Students’ survey responses and self-reported practice hours were not normally distributed; they were characterised by descriptive statistics, such as median, and compared using Mann-Whitney U tests. Students’ likelihood to have practised on models was compared using χ2. The age, weight and body condition score of patients in each group were normally distributed; they were described using mean and SD and were compared using independent samples t-tests. Which week students completed their surgery was compared using a one-way between-subjects analysis of variance. Live surgical rubric item scores were found to be normally distributed; they were described using mean and SD and were compared using an independent samples t-test. Surgical times and global rating scores were not normally distributed; they were described using median and compared using Mann-Whitney U tests. Effect sizes were calculated using Cohen’s d. The P value was set at 0.05, and analyses were performed using SPSS Statistical Package for Social Sciences V.25.
Of the 44 participants, there were 7 males (16 per cent) and 37 females (84 per cent); this gender distribution represented that of the total class, which was 16.3 per cent male and 83.7 per cent female. Of the 29 students randomised to the VR group, 24 (83 per cent) reported using the application. Reasons for not using the application included not being able to use the headset with their phone (1 student), not being able to get the app to work on their phone (1 student) did not want to download the app when other users said the image was blurry and unhelpful (1 student) or gave no reason (2 students). Non-users were excluded from further analysis. The VR group students reported that the VR application was helpful for learning to perform surgery (median=agree). When VR group students were asked if they would recommend the application to others, the responses were relatively evenly distributed from somewhat unlikely to very likely, with a median response of somewhat likely.
Students who used the VR application spent a median of 0.5 hours watching videos in three dimensions with the headset and 1 hour watching in two dimensions with only their smart phone. Seven students (29 per cent) reported using surgical models provided by the school to review and practise their skills prior to live surgery. Among the VR group students who used models in the month prior to their live surgery, 2 hours was the median time spent practising on the models. The total time that VR group students spent preparing to perform live surgery ranged from 1.25 to 19 hours with a median of 5.8 hours (table 2, figure 4).
Of the 15 control group students, 7 (47 per cent) reported practising on models. There was no difference between groups in their likelihood to practise on models (χ2=1.2, P=0.27). Among control group students using models, 2 hours was the median time spent practising on models. The total time spent preparing to perform live surgery ranged from 2 to 18 hours with a median of 5.3 hours. There was no difference in the total time that VR group students and control group students spent in preparation for their first live surgery (P=0.66; table 2, figure 5).
There was no difference in how prepared the groups felt while performing surgery (median=agree for both; P=0.52) or how confident they felt to perform surgery subsequently (median=agree for both; P=1.0). Fourteen of the VR group students (58 per cent) and nine of the control group students (60 per cent) reported feeling anxious while performing surgery; this was similar between groups (median=agree for both; P=0.41). Neither group reported anxiety significant enough to potentially impact the quality of their surgical skills (median=disagree for both; P=0.63; table 3).
Live surgical performance
Raters scored 19 canine ovariohysterectomy surgeries including 7 students from the control group and 12 from the VR group who used the application. Although this was each participant’s first live surgery, students’ baseline surgical skill level gained from their clinical skills programme was compared using the scores on their high-stakes surgical skills assessment the previous semester. There was no difference between baseline skill level of control group students (M=84.4, SD 12.0) and VR group students (M=76.1, SD 13.8; P=0.21, d=0.64, 95 per cent CI 73.3 to 85.1).
Patients had a mean age of 2.0 years (SD 1.5, range 9 months to 5 years), mean weight of 16.2 kg (SD 5.2, range 6–25 kg) and a mean body condition score of 4.4 (SD 0.9, range 3–6) on a 9-point scale. There were no differences between student groups for the patient age, weight or body condition score of patients (P=0.18, 0.79 and 0.18, respectively). Students in each group were similarly divided among each of the three weeks of surgery (F(2,16)=0.73, P=0.93).
Control group students’ ovariohysterectomy rubric scores ranged from 49 to 101 with a mean of 87 (SD 18.1) while VR group students’ scores ranged from 57 to 111 with a mean of 87 (SD 18.8). There was no difference between the groups’ total rubric scores (P=0.93, d=0, 95 per cent CI 79.2 to 95.0). Both groups had global rating scores ranging from 3 to 5. Control group students’ global rating scores had a mean of 4.1 (SD 0.69) while the VR group had a mean of 4.5 (SD 0.67). There was no statistically significant difference between the groups’ global rating scores (P=0.30), though there was a medium effect size (d=0.59).
One control group student’s patient began haemorrhaging during surgery, and a faculty surgeon took over in order to control bleeding. The remaining six control group students performed their surgeries in 88 to 154 minutes with a mean completion time of 107 minutes (SD 23.7). Students in the VR group performed their surgeries in 72 to 156 minutes with a mean completion time of 104 minutes (SD 22.7). There was no difference between groups in the time it took to complete the surgery (P=0.62, d=0.13, 95 per cent CI 95.2 to 115.4).
Researchers in medical education have reported that learning outcomes generally improve when stereoscopic displays such as VR are used in training.19 A recent electroencephalographic study suggests this improvement is the result of greater object recognition.20 However, when we added the VR application to a robust surgical skills curriculum, we found no difference between groups for rubric scores, global rating score or surgical time. Our students had trained repeatedly on an ovariohysterectomy model, and the procedural familiarity that came from practising on the physical model may have diminished the potential benefit offered by VR training.
There was also no difference in the total time that control group and VR group students spent preparing to perform surgery. All students had an identical course load, so perhaps the time students spent was all they had available given the other demands on their time. There was a wide variety in how much time students within each group spent preparing, possibly based on each student’s self-perceived skill level or degree of anxiety about performing surgery. Time spent preparing may also have depended on each student’s relative degree of concern about the didactic courses which take place concurrently with students’ clinical skills training. Preparing for live surgery at the same time as participating in didactic training increases veterinary students’ cognitive load, and additional research would be necessary to understand how students prioritise their study time between both didactic and skills learning. Another possibility is that veterinary students in both groups reached a point where they felt saturated with information. If so, the amount of time to reach saturation appeared similar whether VR was incorporated into preparation or not.
Seventeen per cent of the VR group students never used the application, either never downloading it or citing technical difficulties once they did. If educators want all students to use VR as an educational resource, they will need to provide more proactive technical support. Most students reported that VR training was helpful to their learning. Student feedback also indicated some areas for continued improvement of the application and headsets, such as blurry three-dimensional views and phones going to sleep while in the headsets. These issues may have led students to watch the recordings longer in two dimensions than in three dimensions. Other students reported having to shift positions to keep a surgical site centred within their view. Several students complained that the headsets were too bulky to allow comfortable wear for long, and a few reported that the headset was incompatible with their phones even though the headset’s technical specifications had indicated compatibility. Several students complained that headset viewing gave them a headache or made them feel nauseous. As with many forms of technology, users fall somewhere on a spectrum of use ranging from early adopters to those who accept the technology only once its value is well proven. User feedback is helpful in resolving product challenges and contributing to constant product improvement.
Students spent a median of 90 minutes watching the VR videos. While this may sound brief, a recent study of almost 7 million massive online open course video watching sessions revealed that the median watching time for an instructional video of any length was only 6 minutes.21 Relative to that, 90 minutes of watching the VR videos in this study was a notable undertaking. However, it is unclear whether spending even more time watching the VR videos would have resulted in improved surgical scores. One limitation of the study was that the post-study surveys were performed anonymously in order to encourage candid feedback about the application. Unfortunately, this did not allow us to link the amount of time students used the application to their surgical performance to determine if there may have been a dose-dependent effect on surgical performance. A follow-up study would feature a prescribed use regimen or require students to report their hours of use so that a dose-dependent effect could be investigated.
We had a small sample size for observed surgeries, which is unsurprising when our methods required that each live surgery must be simultaneously overseen by two faculty members: one for student instruction and one for scoring performance for the study. Read et al performed a study with similar methods evaluating canine ovariohysterectomy models with a sample size of 8–10 students per group, which is comparable with our sample size of 7–12 students per group.18 While a larger study would be helpful in evaluating VR technology further, the constrained number of surgical faculty at a given institution makes it challenging to perform larger studies that require observation of student surgeries. One potential solution would be to record student surgeries using a video camera, which has proven reliable in scoring veterinary students’ suturing performance.22 Another limitation of this study was that students were provided with a VR headset that retailed for approximately US$20. A higher-quality headset may have improved the student experience and generated different outcomes. Also, students’ full demographic data were not collected. In retrospect, collecting demographic data such as age may have been helpful to understand which students were using VR training the most. Finally, participants were from a single veterinary school with a strong surgical skills curriculum already in place. Students from schools that lack a clinical skills programme or provide fewer learning resources may have used the VR application more.
In conclusion, minimally interactive VR training was well accepted by most (83 per cent) of veterinary students it was offered to as they prepared for their first live surgery. VR training did not significantly change how much time students spent preparing for surgery overall, potentially because of the rigour of students’ non-surgical course load. Although stereoscopic videos have been shown to present the most benefits to learners for tasks that are depth-related and performed in the near field19—in other words, in fields like surgery—we did not observe a difference in canine ovariohysterectomy performance scores after minimally interactive VR training. A follow-up study with a larger sample size and prescribed use guidelines would be helpful for further evaluating the VR application’s utility for training veterinary students to perform surgery.
The authors would like to acknowledge the support of Dr Jason Johnson and Dr John Dascanio in supporting LMU’s veterinary clinical skills research programme. The authors would also like to acknowlege the influence of surgical skills faculty at Ross University School of Veterinary Medicine for the laboratory sequence listed in Table 1.
Funding This study was funded by Lincoln Memorial University College of Veterinary Medicine. Exero Vet provided the seat licenses for students participating in this study free of charge.
Competing interests None declared.
Ethics approval This study was reviewed and deemed exempt by the Lincoln Memorial University (LMU) Institutional Review Board. Participating dogs were owned by a local animal shelter that provided written consent for their participation.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request. Deidentified participant data can be obtained by emailing the primary author, as approved by IRB.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.