Abstract
BACKGROUND: Despite the utility of three-dimensional (3D) printing in teaching complex anatomy, its application in pelvic ring injury education remains limited. This study aimed to determine whether 3D printed models improve the accuracy of pelvic fracture AO/OTA classification over conventional imaging, and to assess how this effect differs between junior and senior orthopedic residents. METHODS: Imaging data of 10 pelvic ring injury cases were collected to produce corresponding 3D models. Four rounds of classification tests were conducted among orthopedic residents in our department using different imaging modalities: two-dimensional (2D) images alone, 2D images + 3D computed tomography (CT) reconstructions, 2D images + 3D reconstructions + 3D printed models, and 3D printed models alone. The classification accuracy reaching subgroup level and time consumption were analyzed to assess the utility of 3D models. Additionally, a subjective questionnaire was used to gather participants' perceptions of 3D models as educational and diagnostic tools. RESULTS: Among 45 enrolled participants, 18 who completed all rounds were included and grouped by experience: junior (≤ 5 years, n = 10) and senior (> 5 years, n = 8). Compared to using conventional digital imaging, both adjunctive and independent use of 3D models improved classification accuracy in junior group (adjunctive: median difference = 25.00%, 95% CI: 10.00 to 45.00%, Bonferroni-adjusted P = 0.034; independent: median difference = 25.00%, 95% CI: 10.00 to 40.00%, Bonferroni-adjusted P = 0.034). Among senior residents, independent use of 3D models did not improve classification accuracy (median difference = 10.00%, 95% CI: - 10.00 to 15.00%, Bonferroni-adjusted P = 1.000), and adjunctive use significantly reduced accuracy (median difference = 30.00%, 95% CI: 10.00 to 55.00%, Bonferroni-adjusted P = 0.016). Moreover, Both modes of 3D model application significantly increased time consumption relative to conventional imaging (adjunctive: median difference = - 585.50 s, 95% CI: - 1150.01 to - 283.50, Bonferroni-adjusted P < 0.001; independent: median difference = - 211.00 s, 95% CI: - 344.50 to - 82.00, Bonferroni-adjusted P = 0.039). Subjective feedback revealed overall positive attitude toward 3D models, with a stronger preference for their role in educational contexts over clinical decision-making. CONCLUSION: 3D printed models exhibit experience-stratified difference in pelvic fracture classification, improving accuracy for junior residents while reducing it for seniors. Both adjunctive and independent use of 3D models increased time consumption compared to digital imaging alone. Participants predominantly viewed 3D models as classification supplements rather than clinical decision tools, with no consensus on their surgical utility.