Assessing radiographic status of rheumatoid arthritis: introduction of a short erosion scale.

OBJECTIVE

Although radiographs are an important marker of rheumatoid arthritis severity, no valid simple scoring method exists. Current scoring systems are cumbersome, difficult to learn, time consuming, and suitable only for experts. In addition, there is no "gold standard" for radiographic severity, so that it is impossible for the clinician, trialist, or researcher to place patients' scores along a continuum of radiographic severity. We investigated the scoring and scaling properties of radiographs read by the Larsen method, and we developed a shortened scale that can be placed in the perspective of a linear severity continuum.

METHODS

A total of 3,538 paired hand radiographs obtained over a 24 year period were read by Larsen method and evaluated by Rasch analysis. By iterative methods the number of joints was reduced so that proper fitting, scaling, and dimensionality were obtained. The shortened scale was then tested against the full Larsen scale to determine its ability to detect radiographic progression.

RESULTS

A scale consisting of 6 joints on each hand (3 wrist areas and 3 metacarpophalangeal joints) performed as well as the full 28 joint Larsen score, and had improved fit and scaling properties. This Short Erosion Scale (SES) also performed as well as the Larsen score in measuring radiographic progression as measured by comparative effect sizes. In addition, a nomogram was developed to translate SES scores into radiographic severity scores according to a Rasch logit scale.

CONCLUSION

We developed a Short Erosion Scale by reducing the number of joints evaluated in the Larsen method from 28 to 12. The scale is at least as sensitive in detecting change as the full Larsen scale. SES fits the Rasch model and is hierarchical and generally well spaced. Graphs and nomograms are available so that intrinsic radiographic severity can be documented. This scale is suitable for use in clinical practice and observational studies, and may be appropriate for clinical trials as well.