Background/Purpose: in 2010 ACR and EULAR proposed new classification criteria for rheumatoid arthritis (RA). This new set of criteria has been tested in several external populations. The aim of the present study was to summarize the available evidence performing a systematic literature review and meta-analysis of their diagnostic accuracy.

Methods: We searched PubMed, EMBASE, Cochrane and screened the abstracts of the ACR and EULAR congresses from 2010 to 2012. The inclusion criteria were: 1) population of patients with recent onset arthritis, at least one swollen joint, no alternative diagnosis; 2) The ACR/EULAR 2010 criteria (cut-off of 6) as index test; 3) The use of methotrexate (MTX) or disease modifying antirheumatic drugs (DMARDs) as reference standard; 4) Diagnostic accuracy, case control, prospective or retrospective cohort studies; 5) Sufficient data to build a 2×2 table of diagnostic accuracy. Sensitivity (Se) and specificity (Sp) were calculated for each study, data were pooled using a hierarchical summary receiving operator characteristic curve (HSROC) with confidence and prediction intervals. Three separate meta-analysis were performed, considering MTX, DMARDs or their combination as reference standard. To test the robustness of the results, diagnostic odds ratio (DOR) was calculated and an exploratory meta-regression was performed for the analysis of MTX+DMARDs, considering as confounders symptom duration, rheumatoid factor, anti cyclic citrullinated petide antibodies and the timing of assessment of the reference standard. The risk of bias of the included studies was evaluated using the modified version of the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) proposed by the Cochrane collaboration.

Results: a total of 1,257 references were retrieved, after screening title and abstract 4 full papers were included, together with 6 abstracts from the ACR and EULAR congresses. Using MTX as reference standard, the first meta-analysis showed: Se (95% confidence interval, CI) was 0.73 (0.64,0.80), Sp was 0.74 (0.68,0.79), positive likelihood ratio (LR+) (95% CI) was 2.85 (2.53,3.22), negative LR (LR-) 0.35 (0.27,0.45), DOR 8.03 (6.4,10.09). Using DMARDs as reference standard, Se was 0.80 (0.74,0.85), Sp was 0.61 (0.56,0.67), LR+ 2.11 (1.92,2.32), LR- 0.31 (0.25,0.38), DOR 6.74 (5.49,8.28). Using the combination of MTX and DMARDs as reference standard, Se was 0.76 (0.71,0.81), Sp was 0.69 (0.61,0.75), LR+ 2.48 (2.08,2.95), LR- 0.33 (0.29,0.38), OR 7.38 (6.33,8.62). Meta-regression demonstrated no influence of the possible confounders on the results. The risk of bias was low or unclear for most of the studies.

Conclusion: the new classification criteria have a good sensitivity, while specificity is lower. The development of an optimal diagnostic tool for RA is limited by the absence of a real reference standard. In fact, also the decision to start MTX or DMARDs depends on the rheumatologist and the setting. However, these results confirm the mainly classificative and not diagnostic role of the criteria.

ACR Meeting Abstracts - https://acrabstracts.org/abstract/diagnostic-performance-of-the-2010-american-college-of-rheumatologyeuropean-league-against-rheumatism-classification-criteria-for-rheumatoid-arthritis-systematic-literature-review-and-meta-anal/