Objective. To test the reliability and to define the minimal detectabl
e change of a new radiographic scoring method in rheumatoid arthritis
(RA). Methods. Following the recommendations of an expert panel a new
radiographic scoring method was defined. It scores 38 joints [all prox
imal interphalangeal (PIP) and metacarpophalangeal joints, 4 sites in
the wrists, IP of the great toes, and metatarsophalangeals 2 to 5], re
garding only the amount of joint surface destruction on a 0 to 5 scale
for each joint, Each grade represents 20% of joint surface destructio
n. The method was tested by 5 readers on a set of 7 serial radiographs
of hands and forefeet of 7,0 patients with progressive and destructiv
e RA. Analysis of variance was performed, as it provides the best info
rmation about the capability of a method to detect real change and to
define its sensitivity according to the minimal detectable change. Res
ults. Analysis of variance proved a high probability that the readers
found real change with a ratio of intrapatient to intrareader standard
deviation of 2.6. It also confirmed that one reader could detect a ch
ange of 3.5% of the total score with a probability of 95% and that dif
ferent readers agreed upon a change of 4.6%. Inexperienced readers per
formed with comparable results to experienced readers. The time requir
ed for the reading averaged less than 10 minutes for the scoring of on
e set. Conclusion. The new radiographic scoring method proved to be re
liable, precise, and easy to learn, with reasonable cost. Compared to
published data, it may provide better results than the widely used Lar
sen score. These features favor our new method for use in clinical tri
als and in longterm observational studies in RA.