How useful are combinations of blood tests in "rheumatic panels" in diagnosis of rheumatic diseases?

Abstract

Laboratory tests are sometimes combined into "panels," presumably to facilitate a swift and accurate diagnosis. "Rheumatic panels" were available from 16 of 17 members of the American Clinical Laboratory Association. Panels included an average of five tests (range three to 11). Panel prices ranged from $25 to $189. The three tests most common in the available panels were those for rheumatoid factor, antinuclear antibody, and uric acid level. A panel combining these three tests would have a positive predictive value of only 34.6% in identifying rheumatoid arthritis, systemic lupus erythematosus, or gout in a population with joint pain, in which the combined prevalence of these diseases is estimated to be 10%. Therefore, 65.4% of persons with a "positive" test would not have one of these three rheumatic diseases. Lack of independence between diseases and second tests (for example, positive antinuclear antibodies in rheumatoid arthritis) increases misclassification errors. A careful history and physical examination along with serial ordering of a few selected tests appear optimal to establish a clinical diagnosis of a rheumatic disease.