Pediatric Rheumatology Online Journal → June 2003 → Epidemiology, Classification, Immunology and Immunogenetics → Abstract #1

THE INTERNATIONAL LEAGUE OF ASSOCIATIONS FOR RHEUMATOLOGY (ILAR) CRITERIA FOR JUVENILE IDIOPATHIC ARTHRITIS: ARE WE TALKING ABOUT THE SAME THING?

J. E. Weiss,¹ B. S. Gottlieb,¹ N. I. Ilowite,¹ D. R. Flum,² B. A. Eberhard.¹

¹Department of Pediatric Rheumatology, Schneider Children's Hospital, North Shore-Long Island Jewish Health System, New Hyde Park, NY; ²Department of Surgery, University of Washington, Seattle, WA

BACKGROUND: The reproducibility of ILAR criteria among pediatric rheumatologists has not been assessed. The purpose of this study was to determine the degree of inter-observer agreement of ILAR and ACR classifications of patients (pts) with chronic arthritis.
METHODS: Three board certified pediatric rheumatologists, in practice from 7-15 years, were provided clinical information on 25 pts selected randomly out of 256 pts diagnosed with JRA at our institution (32% polyarticular, 56% pauciarticular, 12% systemic). Data provided included age, sex, medical and family history, medications, lab data and examination results from the initial, 6-week and 6-month visits. Observers were given ILAR and ACR criteria and blinded to the pt's name and diagnosis. Kappa statistics were used to assess inter-observer agreement (1=complete agreement, 0 =no agreement).
RESULTS: Pts mean age was 6.4 years +/-5.2, 76% were female. There were 225 observations (3 observers evaluated 25 pts at 3 discrete times). There was moderate agreement among the 3 observers (kappa 0.47, p0.001). Among observers with 10 years or less of clinical experience agreement was higher (77% agreement, kappa 0.71, p0.001). Among all observers there was a moderate level of agreement identified based on disease classification type (kappa for systemic onset 0.08, p=0.3, kappa for pauciarticular 0.23, p=0.001, kappa for polyarticular 0.01, p=0.4. One limitation of this study was that the RF+ polyarticular JRA pts did not have a repeat RF and therefore couldn't be classified as RF+ using ILAR criteria. Kappa did not improve when these pts were removed from analysis. The level of inter-observer agreement at 6 months using the ACR criteria and the ILAR criteria was almost identical with a kappa of 0.61 and 0.64 respectively (p0.001).
CONCLUSION: . Although the goal of ILAR criteria was to create more homogenous groups of pts, there remains variability in the way rheumatologists classify pts.