Serial appraisals of rheumatoid activity by the systemic and articular indexes, performed by seasoned observers on patients who were accustomed to this type of examination, show an interobserver variation no greater than interlaboratory variations in determinations made on stock samples for uric acid, albumin and hemoglobin. However, the variation found when medical students examined new patients was considerably higher. Interobserver variation was not significantly greater than intraobserver variation. The precision and sensitivity of the various items composing the indexes are compared.
Evalutationes serial del activitate rheumatoide a base del indices generalisate e articular, executate per experte observatores in patientes accostumate a iste type de examine, monstra un variation inter le observatores non plus grande que le variation inter laboratorios in le caso de determinationes a base de specimens standard pro acido uric, albumina, e hemoglobina. Tamen, le variation trovate quando studentes de medicina examinava nove patientes esseva considerabilemente plus alte. Le variation inter observatores non esseva significativemente plus grande que le variation intra le corpore de observationes del mesme observator. Es comparate le precision e le sensibilitate del varie com- ponentes in le indices.
0 METHOD FOR EVALUATING the inflammatory activity of rheuma-
N toid arthritis can be acceptable unless its findings are reproducible both in sequential observations by the same observer as well as among different observers. Thus, it should be possible for different observers, examining the same patient, to arrive at an approximately similar numerical value by any given quantitative method. In the case of the systemic and articular indexes,l this kind of reproducibility has been implied by the senior author, but, so far, no serious study of the magnitude of this variation has been published. In this paper we attempt a definitive study of the problem. Some indication as to the validity of several of the items which compose these indexes has been noted by Cohen and Calkins,2 Vaughn et a1.F and Slonim et al.4 Also, the indexes have been utilized in multiobserver drug trials and have yielded statistically valid information in that situation.B-8
SOME PROBLEMS IN STUDY OF INTEROBSERVER VARIATION
It will be recalled that the indexes are derived from six items, three of which are usually classed as "subjective" (measures of stiffness, fatigue, and aspirin From the Connective Tissue Section,