Construct validity of ILAR and EULAR criteria in juvenile idiopathic arthritis: A population based incidence study from the Nordic countries

Objective. New classification criteria (ILAR) have been proposed for juveni le idiopathic arthritis (JIA). They are more descriptive than those formerl y used [American College of Rheumatology (ACR), European League Against Rhe umatism (EULAR)], but require validation against classifications already in use. We validated the ILAR criteria in relation to the EULAR criteria in a prospective, incidence, and population based setting, and analyzed their f easibility. Methods. Construct validity of ILAR and EULAR classification criteria refer s to how closely the 2 instruments are related and how each of them operate s in classifying subgroups/categories. Twenty doctors in 5 Nordic countries collected data from the incidence cases within their catchment areas durin g an 18 month period beginning July 1, 1997. Clinical and serological data from the first year of disease were collected. Results. A total of 322 patients were included. Classification according to the ILAR criteria was possible in 321 patients 290 patients had a disease duration greater than or equal to 3 months and were classified according to the EULAR criteria. One child could only be classified according to the EU LAR criteria. Thus, 31/322 (9.6%) children were classified according to the ILAR criteria only. Forty-eight of 321 (15%) patients did not fit into any category and 6% (20/321) fulfilled criteria for 2 categories. In the ILAR classification 5 out of 7 categories/subgroups have 2 to 5 specified exclus ion criteria that highly discriminate the definition of each patient. In ou r study the exclusion criteria were fulfilled to only a small extent. Conclusion. The EULAR and ILAR criteria differ concerning the operational d efinitions of the subvariables involved, which complicates their comparison . By using ILAR rather than EULAR criteria the number of cases with juvenil e arthritis increased by 10%, considering the first half-year after onset. The validity of the ILAR criteria is low since they often exclude patients from subgroup classification and the possibility of having more than one di agnosis is not negligible. The specified exclusion criteria for some of the subgroups are difficult to fulfill in clinical work and variables involved could be questioned with regard to their consistency.