Sensitivity and specificity of criteria for spondyloarthritis in children with late onset pauciarticular juvenile chronic arthritis as well as their characteristics.

Objective: To evaluate the sensitivity and specificity of criteria designed for spondyloarthritis in a university hospital treated population of children with late onset pauciarticular juvenile chronic arthritis and a control population.

Methods: Four sets of criteria especially designed for juvenile patients: Garmisch-Partenkirchen juvenile spondylitis criteria (= Garmisch), SEA (=seronegative enthesopathy and arthritis) syndrome, Enthesitis Related Arthritis (ERA), Atypical spondyloarthritis for children and two sets of criteria for patients without age specification (European spondyloarthropathy Study Group - ESSG and Amor) were evaluated in a cross-sectional way in a group of 43 consecutive patients with late onset pauciarticular juvenile chronic arthritis (LOPA) seen over a six-month period in the outpatient clinic. These criteria were analysed in 69 patients with other forms of juvenile chronic arthritis as well. The sensitivity and specificity were calculated for each set, as well as positive predictive value and likelihood ratio. The characteristics described in the different sets of criteria were separately evaluated in the LOPA patients and the other patients.

Results: For sensitivity, the Garmisch criteria scored the highest value (97.7%). However, sensitivity was significantly lower in two of the juvenile sets (SEA syndrome and Atypical spondyloarthritis), respectively 44.2% and 51.2%, as opposed to the other criteria (>85%; p<0.01 by Mc Nemar test). Specificity and positive predictive value (PPV) was the highest for the SEA syndrome criteria (98.5%, vs. 95.0%) followed by the ERA (95.6 % vs. 92.1 %) and the Garmisch criteria (94.2% vs. 91.3%). The positive likelihood ratio (LR+) was >10 in SEA (30.5), ERA (18.7) and Garmisch (16.8). The negative likelihood ratio (LR-) was <0.1 only in the Garmisch criteria (0.02).

Conclusion: Sensitivity, specificity, PPV, LR+ and LR- for the Garmisch-Partenkirchen criteria suggest that they classify almost the same population as defined by LOPA. The SEA syndrome criteria, which were not designed to be classification criteria, being very specific, cannot be used in this patient population to classify a sufficient number of patients. The sensitivity and specificity for the ESSG criteria being similar in these children as in adults suggest they have similar characteristics. The Garmisch-Partenkirchen criteria and/or LOPA definition are major candidates for future research in identifying spondyloarthritis in juvenile patients.