Background/Purpose: Since introduction of the new ACR/EULAR classification criteria of rheumatoid arthritis (RA) several studies were published validating the new criteria in different cohorts using different approaches and goldstandards. Performing a systematic literature review we

Want to summerize published studies and assess the sensitivity and specificity of the new ACR/EULAR classification criteria in comparison to the 1987 ACR criteria using different cohorts and goldstandards.

Methods: A systematic literature review was performed in the three main databases (Medline, EMBASE and Cochrane Central). Information of all included studies were extracted and raw data of patients fulfilling different criteria and goldstandards were extracted in order to calculate sensitivity and specificity, positive and negative predictive values as well as standard deviations. Taking into consideration heterogeneity meta-analyses were performed if possible.

Results: In total 1080 articles were retrieved by the search strategy, of which 14 studies (total 6200 patients) could be included.  9 studies included early arthritis patients, 3 established RA, 1 undifferentiated arthritis (UA) and 1 patients with joint symptoms.

5 studies used initiation of MTX (range sensitivity (sens) 0.68 to 0.88 and specificity (spec) 0.3 to 0.72;pooled (95%CI): sens 0.86 (0.84-0.88); spec 0.49 (0.46-0.52)), 3 studies initiation of DMARD (range sens 0.62 to 0.85 and spec 0.38 to 0.78; pooled (95%CI): sens 0.80 (0.78-0.82); spec 0.61 (0.57-0.64)), and 6 studies used expert opinion as goldstandard, out of them 3 in early arthritis patients (range sens 0.62 to 0.91; spec 0.35 to 0.78; pooled (95%CI): sens 0.87 (0.85-0.89) spec 0.45 (0.41-0.49)); 2 in established RA (sens 0.66 to 0.78); 1 in UA (sens 0.47, spec 0.71). Two studies used expert opinion and DMARD as goldstandard (sens 0.58 to 0.74, spec 0.8 to 0.86). Considering different joint counts (JC) no differences of sens and spec could be observed (28JC  3 studies (sens 0.83 to 0.86; spec 0.37 to 0.6); 40JC 3 studies (sens 0.64 to 0.88; spec 0.53 to 0.76);  66/68JC 7 studies (sens 0.48 to 0.88; spec 0.54 to 0.83) no pooling due to different goldstandards/cohorts).

Seven studies directly compared 2010 with 1987 criteria using different goldstandards showing an slightly lower overall specificity (mean delta Sensitivity 2010 Ð Sensitivity 1987 criteria = -0,05) but higher overall sensitivity (mean delta Sensitivity 2010 — Sensitivity 1987 criteria =+0,13) of the new 2010 criteria compared to the 1987 criteria (figure 1)

Conclusion : The new ACR/EULAR classification criteria seems valid independent of goldstandard and cohort used. Compared to the 1987 criteria they show higher sensitivity and almost equal specificity.