Virtual Reality and Simulation in Orthopaedic Training

Virtual reality in surgical training is no longer a conference novelty. It sits beside bench models, cadavers and theatre scenarios as one form of simulation that US and UK orthopaedic curricula now expect programmes to use. The question for a trainee or a programme director is narrower than "does VR work?": which simulator, trained how, has been shown to change what happens in a real operating theatre?

The short answer is that the strongest evidence is in diagnostic arthroscopy, where randomised blinded trials show skills transfer to live cases. For open and trauma procedures, the trials so far stop at the cadaver or the bone model. This article sets out the types of simulator, how validity is judged, what the key trials found, what the curricula require, and where cost and access limit all of it.

Why simulation moved into the curriculum

Two pressures pushed skills training out of the operating theatre. The first is working time. ACGME limits US residents to 80 hours a week, averaged over four weeks, in its orthopaedic surgery programme requirements (opens in a new tab). In the UK, the Working Time Regulations 1998 (opens in a new tab) cap working time at an average of 48 hours for each seven days, unless the worker agrees in writing to opt out. Fewer hours means fewer cases seen during training.

The second is the expectation that a trainee's first attempts at a psychomotor skill should not be on a patient. Arthroscopy shows the problem most clearly: holding the horizon, triangulating instruments and working off a two-dimensional image are skills that observation alone does not teach.

In the winter of 2012 the American Board of Orthopaedic Surgery (ABOS) and the Residency Review Committee agreed changes to PGY-1 training that took effect in July 2013, introducing simulation training with goals, objectives and assessment metrics, and requiring a dedicated space for it. The ABOS account of that decision (opens in a new tab) is explicit that the aim was to move skills training away from an apprenticeship that relied on the operating room.

Types of surgical simulator

"Simulation" covers very different tools. The UK T&O curriculum (opens in a new tab) defines it broadly as any reproduction or approximation of a real event, process or set of conditions, and notes that it does not have to depend on expensive equipment: a suturing aid or a role play can count.

The categories overlap. The trials below show that a bench-top knee model and a virtual reality knee simulator can both produce measurable transfer, so the label matters less than how the training is structured and assessed.

How is a surgical simulator validated?

Most simulator papers use the classical validity vocabulary. It is worth knowing what each term does and does not prove.

Two points matter for reading the literature. First, construct validity is the most often reported and the least useful on its own. Vaghela and colleagues tested the virtual reality version of the FAST arthroscopy programme in 66 surgeons from four training programmes and found that six of eight modules did not discriminate between novices, intermediates and experts; the authors judged the modules unsuitable for assessing arthroscopic proficiency in their present form (Clin Orthop Relat Res 2022 (opens in a new tab)).

Second, education researchers have largely replaced the classical "types" with Messick's unified framework, first proposed in 1989 and adopted as the field standard in 1999 and 2014. It gathers five sources of evidence: content, response process, internal structure, relationships with other variables, and consequences. Cook and Hatala's primer discourages the term "face validity" altogether. A systematic review of 498 surgical simulation validity studies published between 2008 and 2017 found that only 6.6% used the Messick framework, and most still relied on outdated terms such as face validity (Borgersen et al., Ann Surg 2018 (opens in a new tab)).

For practical purposes: when a company or course says a simulator is "validated", ask which kind of evidence it has, and whether anyone measured performance in a real operation.

What do the arthroscopy transfer trials show?

Arthroscopy has the best evidence because the skills are well defined and theatre performance can be recorded and scored blind. The relevant clinical context is in the topic pages on meniscal tears and anterior shoulder instability.

Howells 2008 (Oxford, bench-top knee). Twenty junior orthopaedic trainees were randomised to a fixed protocol of training on a bench-top knee arthroscopy simulator or to no additional training. In theatre, under a blinded consultant, the simulator group scored significantly higher on the Orthopaedic Competence Assessment Project procedure-based assessment (p = 0.0007) and on a global rating scale (p = 0.0011).

Cannon 2014 (seven US centres, VR knee). PGY-3 residents were randomised to train on the ArthroSim virtual reality knee simulator to a proficiency level set by five community orthopaedic surgeons, or to continue usual training. Forty-eight of 54 residents completed the study; the simulator group needed an average of 11 hours to reach proficiency. In a live diagnostic knee arthroscopy scored blind from video, the trained group did better on the procedural checklist (p = 0.031), driven by probing (p = 0.016) rather than visualisation (p = 0.34). The global rating scale narrowly missed significance (p = 0.061), and operating time did not differ, which the authors attributed to less thorough examinations in the control group.

Waterman 2016 (shoulder). Twenty-two trainees were randomised to simulation or standard practice. The simulation group received one hour of instruction in total over three months. On the final recorded diagnostic shoulder arthroscopy, blinded experts rated the simulation group's ASSET safety score higher (3.29 vs 3.00, p = 0.005).

Angelo 2015 (proficiency-based progression, cadaveric Bankart). Forty-four PGY-4 and PGY-5 residents from 21 ACGME programmes were randomised to the standard AANA resident course, the same course with a shoulder model simulator, or proficiency-based progression plus the simulator, in which residents had to reach set benchmarks before moving on. On a blinded three-anchor Bankart repair in a cadaver, the proficiency group made 56% fewer errors than the traditional group and 41% fewer than the simulator-only group. The final proficiency benchmark was reached by 68.7% of the proficiency group, 36.7% of the simulator group and 28.6% of the traditional group. The endpoint was cadaveric, not clinical.

Nicandri 2026 (FAST). In 34 PGY-1 to PGY-3 residents from seven programmes who completed a two-day FAST course, each FAST module passed was associated with a 3.86-point higher ASSET score (0 to 33 scale) and a 4.52-point higher ABOS O-Score (8 to 40 scale) in the operating room. This was a single cohort, so it shows association rather than a randomised effect.

What does the evidence show outside arthroscopy?

The trials in open surgery are smaller and mostly stop short of patients.

These support VR for learning procedural sequence and orientation before a cadaver lab, for operations covered in the topic pages on total hip arthroplasty and tibial shaft fractures. They do not yet show better results for patients.

The systematic reviews say the same. Aïm and colleagues found 10 trials with 303 participants, all in arthroscopy; nine of the ten assessed performance on a simulator, one in the operating room, and none measured a patient-related outcome. Bartlett and colleagues identified 31 studies of orthopaedic VR simulators: 18 showed construct validity of 16 different simulators, eight showed skill acquisition with repeated use, and five showed measurable improvement in operating theatre performance. They concluded that evidence outside knee and shoulder arthroscopy was lacking.

Do simulator skills last?

Not without continued practice. Dunn and colleagues randomised 17 residents to one hour of arthroscopic shoulder simulation or standard practice. After training, the simulation group improved on ASSET score and time to completion. After a year in which neither group used the simulator, there was no difference between the groups on any measure. The authors' conclusion was that programmes need a curriculum that continues through the year, not a one-off session.

That fits the older principles set out by Gallagher and colleagues for integrating VR into surgical training (Ann Surg 2005 (opens in a new tab)): train to an objectively defined proficiency criterion, require consistent performance at that level, and schedule practice at intervals rather than in one massed block.

What do training curricula require?

United States: ACGME and ABOS. The current ACGME orthopaedic surgery requirements (opens in a new tab) say the PGY-1 must include formal instruction in basic surgical skills, delivered longitudinally or as a dedicated rotation. That curriculum must have goals, objectives and assessment metrics; must cover initial management of injured patients (splinting, casting, traction and other immobilisation); and must include basic operative skills: soft-tissue management, suturing, bone management, arthroscopy, fluoroscopy and use of basic orthopaedic equipment. Programmes must also provide a dedicated space for basic surgical skills training.

To help programmes meet the 2013 requirement, an ABOS Surgical Skills Task Force produced 17 PGY-1 skills modules, with low-cost options among its stated goals. ABOS states that the requirements do not mandate this or any other particular curriculum.

The FAST programme. The Fundamentals of Arthroscopic Surgery Training programme, run by the Arthroscopy Association of North America (opens in a new tab), was developed with AAOS, ABOS and Sawbones. It pairs online modules with a physical workstation, used in resident training since 2011, that progresses from a clear shell to an opaque one. The modules cover arthroscope set-up, horizon control, sweeping, telescoping, periscoping, probing, ring transfer, meniscal resection, maze navigation, suture passing and knot tying.

United Kingdom: ISCP. The Trauma and Orthopaedic Surgery curriculum (opens in a new tab) (August 2021, version 2 August 2024) describes simulation as part of a blended approach alongside supervised clinical practice. It states that simulation alone cannot replace supervised practice or authorise a doctor to practise unsupervised, and that feedback and debriefing are integral to it. Assessment still centres on real cases: certification expects an indicative 1,800 operations, and three level 4 procedure-based assessments by two or more trainers in each index operation group. For two groups, supracondylar fracture and external fixator application, one level 4 assessment in a non-simulated setting is acceptable, and the curriculum states that one may be assessed in simulation with the agreement of the assigned educational supervisor and training programme director. Our guide to keeping a surgical logbook covers how that operative evidence is recorded.

Where does augmented reality fit?

Augmented reality overlays digital information on the real field rather than replacing it. In clinical orthopaedics it is used mainly for intraoperative guidance rather than training: the FDA cleared Augmedics' xvision Spine System through the 510(k) route in December 2019, under the device classification "Orthopedic Augmented Reality" (FDA K190929 (opens in a new tab)).

As a training tool the evidence is early. In a randomised trial of 24 final-year medical students orienting an acetabular cup on a phantom pelvis, those trained with live AR feedback ended up as accurate as those taught one-to-one by a hip surgeon (mean difference 1.2°, p = 0.281); 11 of 12 participants said they would prefer an expert trainer for learning combined with AR for unsupervised practice (Logishetty et al., Clin Orthop Relat Res 2019 (opens in a new tab)). Patient-specific rehearsal on printed models is covered in our article on 3D printing in orthopaedic oncology and trauma, and computer-assisted joint replacement in robotic arthroplasty.

What limits simulation: cost, access and realism

Cost has not been studied well. None of the trials above measured cost. Bartlett and colleagues called for analyses of the cost efficiency of validated simulators to justify the investment, and Atesok and colleagues listed unproven feasibility and cost-effectiveness among the main obstacles to building simulation into curricula and board certification (Adv Orthop 2019 (opens in a new tab)). For scale, operating room time in California hospitals in 2014 cost a mean of about $36 to$37 per minute (Childers and Maggard-Gibbons, JAMA Surg 2018 (opens in a new tab)); a simulator that shortened trainee cases would recover some of its price, but no trial here has shown that.

Cadavers are regulated as well as expensive. In the UK, the Human Tissue Authority licenses anatomical examination and the storage of anatomical specimens and of material from the deceased under the Human Tissue Act 2004 (HTA (opens in a new tab)). That restricts cadaveric training to licensed centres and scheduled courses.

Access depends on local funding. High-fidelity simulators are capital purchases that not every department can make. Low-cost options carry much of the load: the ABOS modules listed low-tech, low-cost options among their goals, the ISCP curriculum notes that simulation need not depend on expensive equipment, and the FAST workstation is reusable with no special set-up or clean-up.

Realism has limits. Virtual anatomy behaves as programmed. It does not reproduce unexpected comminution, soft tissue that tears or bleeds, or the ergonomics of positioning a patient and an image intensifier around a theatre team. Scenario-based team simulation covers some of the communication and crisis skills that a desktop simulator leaves out, but only with a structured debrief.

How should a trainee use simulator time?

The trials point to a few practical rules.

• Train to a benchmark, not a clock. Cannon's and Angelo's trials trained residents to a defined proficiency level, and in Angelo's trial that beat the same simulator training without benchmarks. Residents in Cannon's trial needed about 11 hours on average.
• Use the metrics. Time, instrument and camera path length, and camera alignment are among the measures simulators record and the studies scored. Work on the weakest one rather than chasing completion time alone.
• Spread practice out and keep it going. Interval practice is preferred to massed sessions, and a one-hour gain had gone at a year without further training.
• Debrief with a trainer. The ISCP curriculum treats feedback as integral; a trainer can link a metric to its cause, such as portal placement.
• Take the skill into theatre deliberately. Simulation gets you past the first, clumsiest phase of a skill. Judgement in the face of anatomical variation still comes from supervised real cases.