The reliability of the Lane-Sandhu score and the modified RUST for the assessment of postoperative radiographs of long bone non-unions and bone defects.

Abstract

Quantification of bone healing can be of interest for both clinical and research purposes. However, radiographic assessment of bone healing is challenging, especially in postoperative bone defects and non-union. Scores, such as the (modified) Radiographic Union Score for Tibial fractures ((m)RUST) are widely known. The Lane-Sandhu score, a lesser-known score for bone defects, may have benefits over the mRUST score. The aim of this study is to compare the inter- and intraobserver reliability of the Lane-Sandhu score with the mRUST in human non-unions. First, five postoperative radiographs were scored by five observers using both scores. Pitfalls of the scores were thereafter analyzed in a training session. Then, each observer scored ten new radiographs. The intraclass correlation coefficient was calculated to determine the intra- and interobserver reliability of the scores for each session. The pilot session resulted in an interobserver reliability of 0.48 for the mRUST and -0.049 for the Lane-Sandhu score. During the training session, the interobserver reliability scores were 0.50 and 0.14 respectively. The final session resulted in a reliability score of 0.79 (95 % CI 0.60-0.91) for the mRUST and 0.76 (0.59-0.88) for the Lane-Sandhu. Both scores are reliable scoring systems for the interpretation of postoperative bone defects and non-union. There is a slight preference for the mRUST because the reliability was less dependent on the training session. For future research, the interpretation of postoperative radiographs should be well described and a training session is recommended.