The challenge of using the rheumatoid arthritis diagnostic criteria in clinical practice

Abstract

The new 2010 ACR/EULAR (American College of Rheumatology/European League Against Rheumatism) criteria of Rheumatoid Arthritis recently published, have been released to classify and identify patients with early RA who could benefit from early therapy. They recommend anti-citrullinated protein antibody (ACPA) testing as an alternative criterion to Rheumatoid Factor (RF) and ACPA that were introduced together with the other classic criteria in a scoring system. We previously criticized these new criteria because of unavailable specificity and sensibility in the first paper, and the use of ACPA as dichotomous criterion (presence/absent) and alternatives to rheumatoid factor. Our previous work promoted discussion and fostered new research on this issue. By the light of new data, in an effort to improve clinical reasoning, we suggest a more practical probabilistic point of view. In this regard, we analyze the sensitivity and specificity of the diagnostic studies that evaluate the performance of the 2010 classification criteria. Then, we compare the old and the new classification criteria. Subsequently, we describe the use of likelihood ratios applied to the classification criteria and different cutoff levels of ACPA for decision-making in different setting. Moreover, we define some properties of likelihood ratios and their use for diagnosing or excluding rheumatoid arthritis. We want to share this kind of knowledge within the scientific community because we believe that it can help general practitioners and specialists to recognize early arthritis patients implementing a more efficient probabilistic clinical reasoning.