Evaluation of the power of six clustering features in identifying a homogeneous disease subset in juvenile idiopathic arthritis (JIA)

Methods All patients seen in study centers between 1983 and 2004 (N = 750) were classified according to ILAR criteria. Categories of systemic arthritis, RF-positive polyarthritis, and enthesitis related arthritis were excluded because it was felt they represent sufficiently homogeneous entities. Patients in the remaining categories (oligoarthritis persistent and extended, RF-negative polyarthritis, psoriatic arthritis and undifferentiated arthritis) were grouped together (N = 603). In each patient, the presence of the 6 clustering features was assessed. The relative power of each clustering feature in identifying a homogeneous disease subgroup was examined by assessing its ability to separate patients by the presence of the remaining clustering features. Results The ANA revealed the greatest power in separating patients with or without the other clustering features (see table 1), p < 0.01.