ASSESSMENT OF RADIOLOGICAL HEALING RELIABILITY AND ACCURACY IN DISTAL RADIUS FRACTURES

Distal radius fractures (DRFs) are commonly managed non-operatively through closed reduction and cast immobilization; however, complications may arise if the immobilization period is too long or short. There are no standardized guidelines to determine the optimal duration of immobilization, so radiographic assessment of healing progression is common. However, there is no consensus on criteria to assess union. Scoring cortical continuity has been used with some success in lower limb fractures,1 but it remains unclear if a similar technique can reliably determine union after DRFs. In this study, we hypothesized that radiographic assessment based on cortical continuity and fracture line visibility of non-operatively treated DRFs would provide a reliable measure of fracture healing. We assessed inter- and intra-rater reliability of radiographic scoring, with and without access to patient demographics and injury characteristics. We also determined the association of radiographic scoring against Patient-Rated Wrist Evaluation (PRWE) and predictions of bone stiffness. Consecutive adults who underwent non-operative treatment for a DRF were enrolled in this prospective cohort study. Orthogonal radiographs were taken at initial assessment and 2-weeks, 4–6 weeks, and 26-weeks post-fracture. Six independent raters (two attending surgeons, two orthopaedic trauma fellows, and two senior residents) determined radiographic cortex score (RCS) based on callus and fracture line visibility. Raters scored each image thrice, with a minimum two-week washout period between scorings. Patient and injury information was provided at the third scoring. Raters also scored fracture union and whether the cast should be removed (yes/no responses). Intra-class correlation coefficients (ICC) and repeated measures correlation coefficients (rm) were calculated for within and between raters. PRWE was collected at each timepoint, as was a high-resolution peripheral quantitative computed tomography (HR-qQCT) image of the wrist (61 μm isotropic resolution). The latter was used to develop image-based micro-finite element models (μFE) to estimate apparent stiffness under uniaxial compression and torsional loading. Twenty participants (18 female, mean age = 50.5 ± 15.8 years) were enrolled. As shown in Figure 1, the inter-rater agreement was fair at the first and second round, with respective ICCs of 0.69 and 0.68. Inter-rater agreement was stronger at the third round, when patient information was provided (ICC=0.80). Finally, inter-rater agreement for the RCS was consistently stronger than the agreement for union or cast removal. Finally, RCS was strongly associated with μFE-estimated stiffness (rm = 0.74 to 0.84) and PRWE (rm = −0.83 to −0.75). Intra-rater reliability across the first two rounds and rounds one and three were stronger for RCS (mean ICC = 0.69 and 0.73, respectively) than fracture union (ICC = 0.57 and 0.56, respectively) or cast removal (ICC = 0.51 and 0.49, respectively). RCS had fair-to-good inter-rater reliability, across multiple rounds of scoring. Scores were strongly associated with μFE-estimated stiffness and PRWE, suggesting it is a good measure of mechanical and function changes indicative of healing. We observed weaker reliability of more subjective determination of fracture union and the decision to continue cast immobilization. Thus, we recommend the use of the RCS to aid clinical decision-making in non-operatively treated DRFs. For any figures or tables, please contact the authors directly.