Background/Purpose:

Most gout is managed in primary care where the diagnosis seldom relies upon identification of MSU crystals. Several classification criteria for gout have been developed but there is little information on the relative performance of these criteria, especially in comparison to competent identification of MSU crystals in synovial fluid. This study, undertaken as part of an ACR-EULAR project to update gout classification criteria compares the performance of existing criteria.

Methods:

Investigators from 25 sites in 16 countries contributed data on consecutive patients with at least 1 recent swollen joint or subcutaneous nodule that conceivably might be gout. All patients underwent arthrocentesis or tissue aspiration and examination of synovial fluid or tissue with polarizing microscopy by an examiner who had undergone a 2-step certification procedure. Case-control status was defined by the presence or absence of MSU crystals. Data were collected that enabled classification into each of 5 published classification criteria for gout. Both full and survey versions of 1977 American Rheumatism Association (ARA) criteria were used for this analysis. The original NEW YORK and ROME criteria and a modification to contain only clinical items were included. Specificity, sensitivity and AUC were calculated and the differences in test performance were assessed with logistic regression. The reference category was MEX for each regression model.

Results:

Data from 983 patients (509 cases, 474 controls) were collected, of whom 702 were male. The mean (SD) age was 58.5 (17.2) and duration of disease since first ever symptoms was 7.8 (24.6) years. Controls had a clinical diagnosis of gout (MSU crystals not observed by microscopy, n=50), CPPD (109), osteoarthritis (67), rheumatoid arthritis (69), septic arthritis (10), SLE (5), spondyloarthritis (71), undifferentiated arthritis (60), other (31). Not all patients could be classified by every criteria because of missing data, particularly radiographs.

The performance of the criteria is shown in the Table. Analysis of subjects without tophi generally led to minor differences in criteria performance. Excluding controls with a clinical diagnosis of gout led to slightly better specificity (54% MEX to 84% NEW YORK).

Overall, sensitivity was very high. The high sensitivity of the ARA (full) and NEW YORK criteria is due to the fact that presence of MSU crystals is sufficient for classification; therefore all cases meet these criteria by study design. The sensitivity of ARA, ROME and NEW YORK criteria are greatly reduced by excluding MSU crystal data. MEX has worse specificity than all other criteria sets and NETH has worse specificity than NEW YORK or ROME.

Conclusion:

Although sensitivity of existing criteria is satisfactory, the specificity of every set is less than 80%, which is not ideal. There is a need for better performing criteria.