Abstract
To determine whether simplification of the Eaton-Glickel (E-G) classification of trapeziometacarpal (TMC) joint arthrosis (eliminating evaluation of the scaphotrapezial [ST] joint) and information about the patient's symptoms and examination influence interobserver reliability. We also tested the null hypotheses that no patient and/or surgeon factors affect radiographic rating of TMC joint arthrosis and that no surgeon factors affect the radiographic rating of ST joint arthrosis.
In an on-line survey, 92 hand surgeons rated TMC joint arthrosis and ST joint arthrosis separately on 30 radiographs (Robert, true lateral, and oblique views) according to the (modified) E-G classification. We randomly assigned 42 observers to review radiographs alone and also informed 50 of the patient's symptoms and examination. Information about symptoms and examination was randomized. Interobserver reliability was determined with the s* statistic. Because of the hierarchical data structure, cross-classified ordinal multilevel regression analyses were performed to identify factors associated with the severity of arthrosis.
Shortening the E-G classification to the first 3 stages significantly improved the interobserver reliability, which approached substantial agreement. Providing clinical information to observers marginally improved interobserver reliability. Factors associated with a lower E-G stage for TMC joint arthrosis, among observers who rated the severity of TMC joint arthrosis based on radiographs and clinical information, included female surgeon, practice setting, supervising surgical trainees in the operating room, self-reported number of patients with TMC joint arthrosis typically treated annually, male patient, higher patient age, pain limiting daily activities, and shoulder sign. A self-reported larger number of patients with TMC joint arthrosis treated annually was the only variable associated with a higher modified E-G classification to rate ST joint arthrosis.
Our findings suggest that simpler classifications that focus on a single anatomical area are reliable and that surgeon and patient factors can bias interpretation of objective pathophysiology such as radiographic findings.
Diagnostic III.
