[Radiographic signs of scaphoid union after bone grafting: the analysis of inter-observer agreement and intra-observer reproducibility].

INTRODUCTION: The diagnosis of radiological union of scaphoid bone after bone grafting requires clear evidence of bony trabeculae traversing the graft from the proximal to the distal pole on at least two of four standard scaphoid views. This sign is the only objective assessments of union. Radiographs of the scaphoid taken 18 weeks after operation, however, can be difficult to interpret. This fact led us to question whether radiographs of scaphoid at 18 weeks provide reliable and objective indication of union. OBJECTIVE: Our study was, therefore, designed to determine the reliability of the radiographic diagnosis of scaphoid union after bone grafting by testing the degree of inter-observer agreement and reproducibility. METHODS: Out of 30 sets of the scaphoid bone radiographs after bone grafting taken 18 weeks after operation, 15 of good quality were selected. Each set included four views: postero-anterior, lateral, semi- pronated and semi-supinated. Seven observers were tested: three orthopedic consultants, three residents and one consultant in radiology. Each was presented with 15 sets of radiographs designated from 1 to 15 and each was asked to answer the question: "Are there trabeculae crossing the fracture site?" Possible answers were 'yes' or 'no'. Eight weeks later, the same 15 sets of radiographs were marked in alphabetic order from A to K and presented to the same seven observers. Data was then analyzed and expressed in terms of interobserver agreement in pairs and intra-observer reproducibility. Calculation was done by kappa statistics so that the degree of disagreement was taken into account and allowance was made for chance agreement. Kappa values can vary from -1.0 (complete disagreement) through 1 (chance agreement) to +1 (complete agreement). RESULTS: For all 15 sets of radiographs, the degree of agreement between each pair of observers was illustrated in Table 2. It demonstrated the level of agreement between each pairs of seven observers. The overall median kappa coefficient for inter-observer agreement was 0,46. This value corresponded to "moderate" strength of agreement. Median kappa coefficient for consultant was 0.62, but median kappa coefficient for residents was 0.43. The degree of agreement between the opinions of each observer at his first and at his subsequent reviews of the same set of radiograph after 8 weeks was presented in Table 3. The overall median Kappa coefficient for the intraobserver agreement for all seven observers was 0.54. The ability of seven observers to reproduce their own opinion regarding trabeculae crossing the fracture line was poor to use it for assessment of bone union. CONCLUSION: Our conclusion is that radiographs taken 18 weeks after scaphoid grafting cannot be reliable and reproducible for assessment of bone union.