Training in arthroscopy is associated with a steep learning curve for trainees and bears risks for patients. Virtual reality (VR) arthroscopy simulation platforms seek to overcome this and to provide a safe environment for surgical learners. The Fundamentals of Arthroscopic Surgery Training (FAST) program is one such platform. It is currently not known whether the VR FAST program can be employed as a useful teaching or examination tool to assess the basic arthroscopic skills of surgical trainees.
(1) Does the VR FAST program differentiate among novice, intermediate, and expert arthroscopists? (2) Does ambidextrous performance in the VR FAST program correlate with arthroscopic experience?
We prospectively recruited orthopaedic interns (novices), residents (intermediates), and fellows and attendings (experts) to complete the VR FAST program over a 1-year period from four major orthopaedic training programs on a voluntary basis. Sixty-six of 156 invited orthopaedic surgeons participated: 26 of 50 novices (16 men and 10 women), 27 of 65 intermediates (20 men and seven women), and 13 of 41 experts (10 men and three women). Surgeons of any arthroscopic experience were included, with only those with prior experience on the VR FAST program being excluded. The program consists of eight modules: three basic camera modules (Image Centering, Horizon Control, and Telescoping), three advanced camera modules (Periscoping, Trace the Line, and Trace the Curve), and two instrumented bimanual-dexterity modules (Probe Triangulation and Gather the Stars). Time taken to complete each task and measures of economy of movement (camera and instrument path length, camera alignment) were used as measures of arthroscopic experience. Every participant completed the modules using their dominant and nondominant hands. Equality in proficiency in completing the tasks using the dominant and nondominant hands were determined to be measures of arthroscopic experience. Due to the large number of outcome variables, only p values < 0.01 were considered to be statistically significant.
Six of eight VR FAST modules did not discriminate among novice, intermediate, and expert arthroscopy participants. However, two did, and the ones that were most effective at distinguishing participants by level of experience were the Periscoping and Gather the Stars modules. For the Periscoping module using the dominant hand, novices required longer to complete the task with a median time of 231 seconds (IQR 149 to 358) and longer camera path length median of 191 cm (IQR 128 to 273) compared with intermediates who needed 127 seconds (IQR 106 to 233) and 125 cm (IQR 92 to 159) and experts who needed 121 seconds (IQR 93 to 157) and 119 cm (IQR 90 to 134) (p = 0.001 and p = 0.003, respectively). When using the nondominant hand, novices took longer to complete the task with a median time of 231 seconds (IQR 170 to 350) and longer camera path length 204 cm (IQR 169 to 273) compared with intermediates who required 132 seconds (IQR 97 to 162) and 111 cm (IQR 88 to 143) and experts who needed 119 seconds (IQR 104 to 183) and 120 cm (IQR 108 to 166) (p < 0.001 and p < 0.001, respectively). For the Gather the Stars module using the nondominant hand, only the novices needed longer to complete the task at a median of 131 seconds (IQR 112 to 157) and needed a longer grasper path length of 290 cm (IQR 254 to 332) compared with intermediates who needed 84 seconds (IQR 72 to 119) and 232 cm (IQR 195 to 254) and experts who needed 98 seconds (IQR 87 to 107) and 244 cm (IQR 215 to 287) (p < 0.001 and p = 0.001, respectively).
Six of eight VR FAST modules did not demonstrate construct validity, and we found no correlation between arthroscopic experience and ambidextrous performance. Two modules demonstrated construct validity; however, refinement and expansion of the modules is needed with further validation in large prospective trials so that pass-fail thresholds can be set for use in high-stakes examinations.
Most VR FAST modules were not discriminatory; however, they can form essential conceptual and procedural building blocks in an arthroscopic curriculum that are beneficial for novices when developing key psychomotor skills. In their present format, however, they are unsuitable for assessing arthroscopic proficiency.