Use of Generalizability Theory Evaluating Comparative Reliability of the Scapholunate Interval Measurement With X-ray, CT, and US

Abstract

There is no universally accepted standard technique for measuring the scapholunate interval and no specific sources of measurement error have been identified. We aimed to establish a set of normal radiological criteria for the scapholunate interval that could be used in comparative studies of wrist pathology to determine interobserver reliability and to identify potential errors that might influence measurements. A total of 60 healthy volunteers participated in the study. Scapholunate interval was measured by three independent observers using X-ray, CT, and US in four positions, including neutral, fist, radial and ulnar deviation. Inter-observer reliability was tested using intraclass correlation coefficient. Generalizability theory was applied to evaluate specific sources of measurement error related to participant, observer, imaging modality and measurement position. In neutral position, the scapholunate interval measured by X-ray, CT, and US was 3.1 mm, 3.5 mm and 3.5 mm respectively. The interval remained constant during fist and radial deviation but decreased during ulnar deviation. Correlation coefficients ranged from 0.874 to 0.907 for X-ray, 0.773-0.881 for CT and 0.964-0.979 for US. In the generalizability study, the participant x modality x position interaction accounted for the largest proportion of total variance (29%). X-ray, CT, and US are reliable modalities for measuring the scapholunate interval, with US having the highest reliability. Participant and position factors may independently contribute to measurement error. Therefore, standardized measurement positions are recommended to obtain reliable measurement results.