Wollheim F A, Pettersson H, Saxne T, Sjöblom K G
Department of Medicine, Malmö General Hospital, Sweden.
Scand J Rheumatol. 1988;17(6):445-53. doi: 10.3109/03009748809098805.
Radiographs of hands and feet were obtained from 125 consecutive patients with rheumatoid arthritis (RA) and the degree of destruction was assessed numerically on a 200-point scale using Larsen's standard radiographs as reference. The method is shown to possess a satisfactory degree of reproducibility. In 96 of these 125 patients, values of another 15 simultaneously determined clinical and biochemical variables were obtained. On applying linear and quadratic multiple regression analysis to this set as well as to the male and female subsets, an 'automatic' selection procedure (stepwise regression) proved duration of disease to be the most important factor relating to the 'Larsen index'. The 96 patients were therefore ranked with respect to duration of disease and divided into 4 subsets of equal magnitude. In the 3 subsets with duration of disease less than 21 years, stepwise regression produced in the final step linear or quadratic combinations not containing duration of disease but correlating quite well with the 'Larsen index' (R = 0.64-0.96). A similar result was obtained upon performing an analogous procedure in the female subset. In all instances, positive contributions of varying degree were obtained from Ritchie's index, ESR, a-antitrypsin (A1-AT), orosomucoid, fibrinogen, and IgM, while negative correlations were associated with ceruloplasmin, IgG, and IgA.
