To evaluate the performance of the American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) 2010 classification criteria for rheumatoid arthritis (RA) with a systematic literature review and a meta-analysis.
PubMed, Embase, Cochrane Library and the abstracts of the ACR and EULAR meetings (2010–2012) were searched for original articles or abstracts with the following inclusion criteria: 1) recent onset arthritis, with at least one swollen joint and no alternative diagnosis; 2) the ACR/EULAR 2010 criteria as index test; 3) the prescription of methotrexate (MTX) or disease modifying antirheumatic drugs (DMARDs) at any time during follow-up as reference standard. Data were pooled using the bivariate model. Three meta-analyses were performed with MTX (primary analysis), DMARDs or their combination (secondary analyses) as reference standard. Heterogeneity was formally tested and explored performing an influence analysis.
The search identified 1,277 references. Six full papers and 4 abstracts met the inclusion criteria. With MTX as reference standard, sensitivity (95% confidence interval, CI) was 0.80 (0.74,0.85), specificity 0.61 (0.56,0.67), positive likelihood ratio (LR) 2.11 (1.92,2.32), negative LR 0.31 (0.25,0.38) and the diagnostic odds ratio (DOR) was 6.74 (5.49,8.28). Using DMARDs as reference standard, sensitivity was 0.73 (0.64,0.80), specificity was 0.74 (0.68,0.79), LR+2.85 (2.53,3.22), LR− 0.35 (0.27,0.45) and DOR 8.03 (6.4,10.09). Using the combination of MTX and DMARDs as reference standard, intermediate results were obtained. The influence analysis detected one potentially influential study. However, its exclusion from the meta-analysis did not have a clinically relevant impact on the results.
The new classification criteria have good sensitivity, lower specificity and an overall moderate diagnostic accuracy. These results confirm that the criteria have classificative and not diagnostic function.