Reliability of radiological assessment following total elbow arthroplasty: a varying observer agreement

BACKGROUND: Total elbow arthroplasty (TEA) is used to treat a variety of elbow pathologies. Survival rates following TEA remain relatively low and complications are often reported. Radiographic evaluation has been used to detect complications. However, standardized assessment for evaluation is missing and reliability of investigating different parameters has never been investigated. Therefore, this study aimed to evaluate intra- and interobserver reliability of the most commonly used radiological parameters following TEA.

METHODS: Anteroposterior and lateral radiographs of 29 patients who underwent Latitude TEA were evaluated for 14 parameters by four experienced upper limb orthopedic surgeons. Primary outcome parameter (implant positioning), parameters with direct clinical consequences (radial head disengagement, periprosthetic fractures, shift, subluxation/dislocation, implant breakage), and parameters with long-term clinical consequences (vertical cementation index, heterotopic ossification, radiolucent lines, bone resorption, center of rotation, metallosis, wear) were evaluated. Intra- and interobserver agreement was determined by intraclass correlation coefficient (ICC) for continuous and Fleiss kappa (k) for categorical parameters. ICC > 0.75 and k > 0.81 were considered reliable.

RESULTS: Determining implant positioning and vertical cementation index appeared reliable for both intra- and interobserver agreement. The other outcome parameters showed much variability, with agreements varying from poor to almost perfect but overall appeared insufficiently reliable.

CONCLUSION: The results of this study demonstrate that implant positioning and vertical cementation index in radiographs of total elbow prostheses can be determined reliably and can be used to assess the postoperative result. The other outcome parameters - e.g., radial head disengagement, periprosthetic fractures, shift, dislocation, heterotopic ossifications, radiolucencies, bone resorption, center of rotation, metallosis, wear - appeared overall insufficiently reliable and should not yet be used until more research into definition, measurement, and scoring of these parameters is conducted. Also, more education for detecting these parameters on radiographs is essential before a standardized assessment protocol can be created.