Sensitivity and specificity of post-operative interference gap assessment on plain radiographs after cementless primary THA
Belt, M., Gliese, B., Muharemovic, O., Malchau, H., Husted, H., Troelsen, A. & Gromov, K. 28 dec. 2018 I : Magnetic Resonance Imaging. 54, s. 103-107 5 s.
Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › peer review
INTRODUCTION: Implant performance of cementless THA is often evaluated by radiolucency on plain radiographs, often classified as interference gaps on direct post-operative radiographs. However, the diagnostic performance is unknown. The aim was to evaluate the diagnostic performance of radiographic assessment of post-operative gaps after primary THA by comparing it with CT confirmed gaps, and secondary to define optimal cut-off criteria for assessing gaps on plain radiographs compared with CT.
MATERIAL AND METHODS: Patients (N = 40) with a primary cementless THA performed between July 2015 and March 2016 were enrolled in the study. Radiolucency was assessed on post-operative AP pelvic digital radiographs by two observers independently. Maximum width and percentage of coverage per zone were reported. Gap volume was measured by manual segmentation on CT images.
RESULTS: When defining a gap as a radiolucency extending through >50% of a zone, the interrater agreement Kappa was 0.241. Sensitivity was 65.8% for observer 1 (Kappa = 0.432), and 86.8% for observer 2 (Kappa = 0.383). When defining a gap as a radiolucency with a width >1 mm, the interrater agreement Kappa was 0.302. Sensitivity was 55.3% and 50% for observer 1 and observer 2, respectively. The ROC-curve resulted in an optimal threshold of 0.65 mm (AUROC = 0.888) and 0.31 mm (AUROC = 0.961) for the two observers.
CONCLUSION: The diagnostic performance of observers detecting interference gaps on radiographs showed low sensitivity. Further on, the inter-rater agreement is too low to do a general recommendation about thresholds for defining gaps. Evaluating progression of radiolucency on radiographs should be performed in the light of these findings.
Originalsprog | Engelsk |
---|---|
Tidsskrift | Magnetic Resonance Imaging |
Vol/bind | 54 |
Sider (fra-til) | 103-107 |
Antal sider | 5 |
ISSN | 0961-9275 |
DOI | |
Status | E-pub ahead of print - 28 dec. 2018 |