Estimation of the area of skin involvement is an important part of atopic e
czema severity scoring systems, We examine the variation seen when six derm
atologically trained observers assessed disease extent in six patients with
atopic eczema using the rule of nines, The median scores ranged from 4.8%
to 37.2%. The level of agreement for the classification of scores into quin
tile categories was 55%, with a chance-corrected agreement (kappa statistic
) of 0.09, representing very poor interobserver agreement. To measure the e
ffects of health care intervention accurately, reliable methods of assessme
nt should be used. Scoring methods in which reliability cannot be demonstra
ted may be misleading and should be interpreted with caution.